Broadband Modem Mixed Signal Front End

AD9866

Rev. 0

Information furnished by Analog Devices is believed to be accurate and reliable.
However, no responsibility is assumed by Analog Devices for its use, nor for any
infringements of patents or other rights of third parties that may result from its use.
Specifications subject to change without notice. No license is granted by implication
or otherwise under any patent or patent rights of Analog Devices. Trademarks and
registered trademarks are the property of their respective owners.

One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A.
Tel: 781.329.4700

www.analog.com

Fax: 781.326.8703

FEATURES

Low cost 3.3 V CMOS MxFE

for broadband modems

12-bit D/A converter

2Ч/4Ч interpolation filter
200 MSPS DAC update rate

Integrated 23 dBm line driver with 19.5 dB gain control
12-bit, 80 MSPS A/D converter
-12 dB to +48 dB low noise RxPGA (< 2.5 nV/rtHz)
Third order programmable low-pass filter
Flexible digital data path interface

Half- and full-duplex operation
Backward compatible with AD9975 and AD9876

Various power-down/reduction modes
Internal clock multiplier (PLL)
2 auxiliary programmable clock outputs
Available in 64-lead chip scale package or bare die

APPLICATIONS

Powerline networking
VDSL and HPNA

FUNCTIONAL BLOCK DIAGRAM

XTAL

RX

AD9866

0 TO 7.5dB

04560-0-001

0 TO 12dB

CLK
SYN.

ADC

80MSPS

2-4X

IOUT_G+
IOUT_N+
IOUT_N
IOUT_G

CLKOUT_1
CLKOUT_2

OSCIN

RX+

IAMP

TxDAC

IOU

_P+

IOU

CLK

MULTIPLIER

2-POLE

LPF

1-POLE

LPF

0 TO 6dB

= 1dB

 6 TO 18dB

= 6dB

6 TO 24dB

= 6dB

SPI

AGC[5:0]

RXCLK

RXE/SYNC

ADIO[11:6]/

Tx[5:0]

ADIO[5:0]/

Rx[5:0]

TXCLK

TXEN/SYNC

MODE

PWR DWN

Figure 1.

GENERAL DESCRIPTION

The AD9866 is a mixed-signal front end (MxFE) IC for
transceiver applications requiring Tx and Rx path functionality
with data rates up to 80 MSPS. Its flexible digital interface,
power saving modes, and high Tx-to-Rx isolation make it well
suited for half- and full-duplex applications. The digital inter-
face is extremely flexible allowing simple interfaces to digital
back ends that support half- or full-duplex data transfers, thus
often allowing the AD9866 to replace discrete ADC and DAC
solutions. Power saving modes include the ability to reduce
power consumption of individual functional blocks or to power
down unused blocks in half-duplex applications. A serial port
interface (SPI®) allows software programming of the various
functional blocks. An on-chip PLL clock multiplier and
synthesizer provide all the required internal clocks, as well as
two external clocks from a single crystal or clock source.

The Tx signal path consists of a bypassable 2Ч/4Ч low-pass
interpolation filter, a 12-bit TxDAC, and a line driver. The
transmit path signal bandwidth can be as high as 34 MHz at an
input data rate of 80 MSPS. The TxDAC provides differential
current outputs that can be steered directly to an external load

or to an internal low distortion current amplifier. The current
amplifier (IAMP) can be configured as a current or voltage
mode line driver (with two external npn transistors) capable of
delivering in excess of 23 dBm peak signal power. Tx power can
be digitally controlled over a 19.5 dB range in 0.5 dB steps.

The receive path consists of a programmable amplifier
(RxPGA), a tunable low pass filter (LPF), and a 12-bit ADC. The
low noise RxPGA has a programmable gain range of -12 dB to
+48 dB in 1 dB steps. Its input referred noise is less than
3.3 nV/rtHz for gain settings beyond 30 dB. The receive path
LPF cutoff frequency can either be set over a 15 MHz to
35 MHz range or simply bypassed. The 12-bit ADC achieves
excellent dynamic performance over a 5 MSPS to 80 MSPS
span. Both the RxPGA and the ADC offer scalable power
consumption allowing power/performance optimization.

The AD9866 provides a highly integrated solution for many
broadband modems. It is available in a space-saving 64-lead
chip scale package and is specified over the commercial (-40°C
to +85°C) temperature range.

AD9866

Rev. 0 | Page 2 of 48

TABLE OF CONTENTS

Specifications..................................................................................... 3

Tx Path Specifications.................................................................. 3

Rx Path Specifications.................................................................. 4

Power Supply Specifications ....................................................... 5

Digital Specifications ................................................................... 6

Serial Port Timing Specifications............................................... 7

Half-Duplex Data Interface (ADIO Port) Timing
Specifications ................................................................................ 7

Full-Duplex Data Interface (Tx and Rx PORT) Timing
Specifications ................................................................................ 8

Absolute Maximum Ratings............................................................ 9

Thermal Characteristics .............................................................. 9

ESD Caution.................................................................................. 9

Pin Configuration and Function Descriptions........................... 10

Typical Performance Characteristics ........................................... 12

Rx Path Typical Performance Characteristics ........................ 12

TxDAC Path Typical Performance Characteristics ............... 16

IAMP Path Typical Performance Characteristics .................. 18

Serial Port ........................................................................................ 19

Serial Port Interface (SPI) ......................................................... 21

Digital Interface .............................................................................. 23

Half-Duplex Mode ..................................................................... 23

Full-Duplex Mode ...................................................................... 24

RxPGA Control .......................................................................... 25

TxPGA Control .......................................................................... 27

Transmit Path .................................................................................. 28

Digital Interpolation Filters ...................................................... 28

TxDAC and IAMP Architecture............................................... 28

Tx Programmable Gain Control .............................................. 30

TxDAC Output Operation........................................................ 30

IAMP Current Mode Operation .............................................. 30

IAMP Voltage Mode Operation ............................................... 31

IAMP Current Consumption Considerations........................ 32

Receive Path .................................................................................... 33

Rx Programmable Gain Amplifier........................................... 33

Low-Pass Filter ........................................................................... 34

Analog to Digital Converter (ADC)........................................ 35

AGC Timing Considerations.................................................... 36

Clock Synthesizer ........................................................................... 37

Power Control and Dissipation .................................................... 39

Power-Down ............................................................................... 39

Half-Duplex Power Savings ...................................................... 39

Power Reduction Options......................................................... 40

Power Dissipation ...................................................................... 42

Mode Select upon Power-Up and Reset.................................. 42

Analog and Digital Loop-back Test Modes ............................ 43

PCB Design Considerations.......................................................... 44

Component Placement.............................................................. 44

Power Planes and Decoupling .................................................. 44

Ground Planes ............................................................................ 44

Signal Routing ............................................................................ 44

Evaluation Board ............................................................................ 46

Outline Dimensions ....................................................................... 47

Ordering Guide .......................................................................... 47

REVISION HISTORY

Revision 0: Initial Version

AD9866

Rev. 0 | Page 3 of 48

SPECIFICATIONS

TX PATH SPECIFICATIONS

Table 1. AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; f

OSCIN

= 50 MHz, f

DAC

= 200 MHz, R

SET

= 2.0 k, unless

otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

TxDAC DC CHARACTERISTICS

Resolution Full

Bits

Update Rate

Full

200

MSPS

Full-Scale Output Current (IOUTP_FS)

Full

Gain Error

25°C I

±2

Offset Error

25°C

Voltage Compliance Range

Full

-1

+1.5

TxDAC GAIN CONTROL CHARACTERISTICS

Minimum Gain

25°C

-7.5

Maximum Gain

25°C

Gain Step Size

25°C

0.5

Gain Step Accuracy

25°C

Monotonic

Gain Range Error

25°C

±2

TxDAC AC CHARACTERISTICS

Fundamental

0.5

dBm

Signal-to-Noise and Distortion

Full

66.6

69.2

dBc

Signal-to-Noise Ratio

Full

68.4

69.8

dBc

THD Full

-79

-68.7

dBc

SFDR Full

68.5

dBc

IAMP DC CHARACTERISTICS

IOUTN Full-Scale Current = IOUTN+ + IOUTN-

Full

105

IOUTG Full-Scale Current = IOUTG+ + IOUTG-

Full

150

AC Voltage Compliance Range

Full

IAMPN AC CHARACTERISTICS

Fundamental 25°C

dBm

IOUTN SFDR (Third Harmonic)

Full

43.3

45.2

dBc

IAMP GAIN CONTROL CHARACTERISTICS

Minimum Gain

25°C

-19.5

Maximum Gain

25°C

Gain Step Size

25°C

0.5

Gain Step Accuracy

25°C

Monotonic

IOUTN Gain Range Error

25°C

0.5

REFERENCE

Internal Reference Voltage

25°C

1.23

Reference Error

Full

0.7

3.4

Reference Drift

Full

ppm/

Tx DIGITAL FILTER CHARACTERISTICS (2Ч INTERPOLATION)

Latency (Relative to 1/ F

DAC

) Full

Cycles

-0.2 dB Bandwidth

Full

0.2187

OUT

DAC

-3 dB Bandwidth

Full

0.2405

OUT

DAC

Stop-Band Rejection (0.289 F

DAC

to 0.711 F

DAC

) Full

Tx DIGITAL FILTER CHARACTERISTICS (4Ч Interpolation)

Latency (Relative to 1/ F

DAC

) Full

Cycles

-0.2 dB Bandwidth

Full

0.1095

OUT

DAC

AD9866

Rev. 0 | Page 4 of 48

Parameter Temp

Test

Level

Min

Typ

Max

Unit

-3 dB Bandwidth

Full

0.1202

OUT

DAC

Stop Band Rejection (0.289 f

OSCIN

to 0.711 f

OSCIN

) Full

PLL CLK MULTIPLIER

OSCIN Frequency Range

Full

MHz

Internal VCO Frequency Range

Full

200

MHz

Duty Cycle

Full

OSCIN Impedance

25°C

100//3

//pF

CLKOUT1 Jitter

25°C III

rms

CLKOUT2 Jitter

25°C III

rms

CLKOUT1 and CLKOUT2 Duty Cycle

Full III

55 %

Gain error and gain temperature coefficients are based on the ADC only (with a fixed 1.23 V external reference and a 1 V p-p differential analog input).

TxDAC IOUTFS = 20 mA, differential output with 1:1 transformer with source and load termination of 50 , F

OUT

= 5 MHz, 4Ч interpolation.

IOUN full-scale current = 80 mA, f

OSCIN

= 80 MHz, f

DAC

=160 MHz, 2Ч interpolation.

Use external amplifier to drive additional load.

Internal VCO operates at 200 MHz , set to divide-by

-1.

Because CLKOUT2 is a divided down version of OSCIN, its jitter is typically equal to OSCIN.

CLKOUT2 is an inverted replica of OSCIN, if set to divide-by-1.

RX PATH SPECIFICATIONS

Table 2. AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; half- or full-duplex operation with CONFIG = 0 default
power bias settings, unless otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

Rx INPUT CHARACTERISTICS

Input Voltage Span (RxPGA gain = -10 dB)

Full

III

6.33

V p-p

Input Voltage Span (RxPGA gain = +48 dB)

Full

III

mV p-p

Input Common-Mode Voltage

25°C

III

1.3

Differential Input Impedance

25°C

III

400
4.0

Input Bandwidth (with RxLPF Disabled, RxPGA = 0 dB)

25°C

III

MHz

Input Voltage Noise Density (RxPGA Gain = 36 dB, f

-3 dBF

= 26 MHz)

25°C

III

2.7

nV/rtHz

Input Voltage Noise Density (RxPGA Gain = 48 dB, f

-3 dBF

= 26 MHz)

25°C

III

2.4

nV/rtHz

RxPGA

CHARACTERISTICS

Minimum Gain

25°C

III

-12

Maximum Gain

25°C

III

Gain Step Size

25°C

III

Gain Step Accuracy

25°C

III

Monotonic

Gain Range Error

25°C

III

0.5

RxLPF

CHARACTERISTICS

Cutoff Frequency (f

-3 dBF

) range

Full

III

MHz

Attenuation at 55.2 MHz with f

-3 dBF

= 21 MHz

25°C

III

Pass-Band Ripple

25°C

III

±1

Settling Time to 5 dB RxPGA Gain Step @ f

ADC

= 50 MSPS

25°C

III

Settling Time to 60 dB RxPGA Gain Step @ f

ADC

= 50 MSPS

25°C

III

100

ADC DC CHARACTERISTICS

Resolution

Bits

Conversion Rate

FULL

MSPS

RX PATH LATENCY

Full-Duplex Interface

Full

10.5

Cycles

Half-Duplex Interface

Full V

10.0

Cycles

AD9866

Rev. 0 | Page 5 of 48

Parameter Temp

Test

Level

Min

Typ

Max

Unit

Rx PATH COMPOSITE AC PERFORMANCE @ f

ADC

= 50 MSPS

RxPGA Gain = 48 dB (Full-Scale = 8.0 mV p-p)

Signal-to-Noise and Distortion (SNR)

25°C

III

43.7

dBc

Total Harmonic Distortion (THD)

25°C

III

-71

dBc

RxPGA Gain = 24 dB (Full-Scale = 126 mV p-p)

Signal-to-Noise (SNR)

25°C

III

63.1

dBc

Total Harmonic Distortion (THD)

25°C

III

-67.2

dBc

RxPGA Gain = 0 dB (Full-Scale = 2.0 V p-p)

Signal-to-Noise and Distortion (SINAD)

Full

64.3

dBc

Total Harmonic Distortion (THD)

Full

-67.3

dBc

Rx PATH COMPOSITE AC PERFORMANCE @ f

ADC

= 80 MSPS

RxPGA Gain = 48 dB (Full-Scale = 8.0 m V p-p)

Signal-to-Noise (SNR)

25°C

III

41.8

dBc

Total Harmonic Distortion (THD)

25°C

III

-67

dBc

RxPGA Gain = 24 dB (Full-Scale = 126 m V p-p)

Signal-to-Noise (SNR)

25°C

III

58.6

dBc

Total Harmonic Distortion (THD)

25°C

III

-62.9

dBc

RxPGA Gain = 0 dB (Full-Scale = 2.0 V p-p)

Signal-to-Noise (SNR)

25°C

61.1 62.9

dBc

Total Harmonic Distortion (THD)

25°C

-70.8

-60.8

dBc

Rx-to-Tx PATH FULL-DUPLEX ISOLATION
(1 V p-p, 10 MHz Sine Wave Tx Output)

RxPGA Gain = 40 dB

IOUTP± Pins to RX± Pins

25°C III

dBc

IOUTG± Pins to RX± Pins

25°C III

dBc

RxPGA Gain = 0 dB

IOUTP± Pins to RX± Pins

25°C III

123

dBc

IOUTG± Pins to RX± Pins

25°C

III

dBc

Includes RxPGA, ADC pipeline, and ADIO bus delay relative to f

ADC

= 5 MHz, AIN = -1.0 dBFS , LPF cut-off frequency set to 15.5 MHz with Reg. 0x08 = 0x80.

= 5 MHz, AIN = -1.0 dBFS , LPF cut-off frequency set to 26 MHz with Reg. 0x08 = 0x80.

POWER SUPPLY SPECIFICATIONS

Table 3. AVDD = 3.3 V, DVDD = CLKVDD = DRVDD = 3.3 V; R

SET

= 2 k, full-duplex operation with f

DATA

= 80 MSPS,

unless

otherwise noted

Parameter

Temp

Test Level

Min

Typ

Max

Unit

SUPPLY

VOLTAGES

AVDD

Full

3.135

3.3

3.465

CLKVDD

Full

3.0

3.3

3.6

DVDD

Full

3.0

3.3

3.6

DRVDD

Full

3.0

3.3

3.6

IS_TOTAL (Total Supply Current)

Full

406

475

POWER CONSUMPTION

AVDD

+ I

CLKVDD

(Analog Supply Current)

311 342 mA

DVDD

+ I

DRVDD

(Digital Supply Current)

Full

133

POWER CONSUMPTION (Half-Duplex Operation with f

DATA

= 50 MSPS)

Mode

AVDD

+ I

CLKVDD

25°C

112

130

DVDD

+ I

DRVDD

25°C IV

49.5 mA

AD9866

Rev. 0 | Page 6 of 48

Parameter

Temp

Test Level

Min

Typ

Max

Unit

Mode

AVDD

+ I

CLKVDD

25°C

225 253 mA

DVDD

+ I

DRVDD

25°C

36.5

POWER CONSUMPTION OF FUNCTIONAL BLOCKS

AVDD

+ I

CLKVDD

)

RxPGA and LPF

25°C

III

ADC 25°C

III

108

TxDAC

25°C

III

IAMP (Programmable)

25°C

III

120

Reference

25°C

III

170

CLK PLL and Synthesizer

25°C

III

107

MAXIMUM ALLOWABLE POWER DISSIPATION

Full

1.66

STANDBY POWER CONSUMPTION

IS_TOTAL (Total Supply Current)

Full

POWER DOWN DELAY (USING PWR_DWN PIN)

RxPGA and LPF

25°C

III

440

ADC

25°C

III

TxDAC

25°C

III

IAMP

25°C

III

CLK PLL and Synthesizer

25°C

III

POWER UP DELAY (USING PWR_DWN PIN)

RxPGA and LPF

25°C

III

7.8

µs

ADC

25°C

III

TxDAC

25°C

III

µs

IAMP

25°C

III

CLK PLL and Synthesizer

25°C

III

µs

Default power-up settings for MODE = HIGH and CONFIG = LOW, IOUTP_FS = 20 mA, does not include IAMP's current consumption, which is application dependent.

Default power-up settings for MODE = LOW and CONFIG = LOW .

DIGITAL SPECIFICATIONS

Table 4. AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; R

SET

= 2 k, unless otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

CMOS LOGIC INPUTS

High Level Input Voltage

Full

DRVDD 0.7

Low Level Input Voltage

Full

0.4

Input Leakage Current

µA

Input Capacitance

Full

CMOS LOGIC OUTPUTS (C

LOAD

= 5 pF)

High Level Output Voltage (I

= 1 mA)

Full

DRVDD 0.7

Low Level Output Voltage (I

= 1 mA)

Full

1.2

Output Rise/Fall Time (High Strength Mode and C

LOAD

= 15 pF)

Full

1.5/2.3

Output Rise/Fall Time (Low Strength Mode and C

LOAD

= 15 pF)

Full

1.9/2.7

Output Rise/Fall Time (High Strength Mode and C

LOAD

= 5 pF)

Full

0.7/0.7

Output Rise/Fall Time (Low Strength Mode and C

LOAD

= 5 pF)

Full

1.0/1.0

RESET

Minimum Low Pulse Width (Relative to f

ADC

)

Clock
cycles

AD9866

Rev. 0 | Page 7 of 48

SERIAL PORT TIMING SPECIFICATIONS

Table 5. AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%, unless otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

WRITE OPERATION (See Figure 46)

SCLK Clock Rate (f

SCLK

)

Full IV

MHz

SCLK Clock High (t

)

Full IV

SCLK Clock Low (t

LOW

)

Full IV

SDIO to SCLK Setup Time (t

)

Full IV

SCLK to SDIO Hold Time (t

)

Full IV

SEN to SCLK Setup Time (t

)

Full IV

SCLK to SEN Hold Time (t

)

Full IV

READ OPERATION (See Figure 47 and Figure 48)

SCLK Clock Rate (f

SCLK

)

Full IV

MHz

SCLK Clock High (t

)

Full IV

SCLK Clock Low (t

LOW

)

Full IV

SDIO to SCLK Setup Time (t

)

Full IV

SCLK to SDIO Hold Time (t

)

Full IV

SCLK to SDIO (or SDO) Data Valid Time (t

)

Full IV

SEN to SDIO Output Valid to Hi-Z (t

)

Full IV

HALF-DUPLEX DATA INTERFACE (ADIO PORT) TIMING SPECIFICATIONS

Table 6. AVDD = 3.3 V ±5%, DVDD = CLKVDD = DRVDD = 3.3 V ±10%, unless otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

READ OPERATION (See Figure 50)

Output Data Rate

Full II

5 80 MSPS

Three-State Output Enable Time (t

PZL

)

Full II

Three-State Output Disable Time (t

PLZ

)

Full II

Rx Data Valid Time (t

)

Full II

Rx Data Output Delay (t

)

Full II

WRITE OPERATION (See Figure 49)

Input Data Rate (1Ч Interpolation)

Full

MSPS

Input Data Rate (2Ч Interpolation)

Full

MSPS

Input Data Rate (4Ч Interpolation)

Full

MSPS

Tx Data Setup Time (t

)

Full II

12.5

Tx Data Hold Time (t

)

Full II

Latch Enable Time (t

)

Full II

Latch Disable Time (t

DIS

)

Full II

AD9866

Rev. 0 | Page 8 of 48

FULL-DUPLEX DATA INTERFACE (Tx AND Rx PORT) TIMING SPECIFICATIONS

Table 7. AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%, unless otherwise noted

Parameter Temp

Test

Level

Min

Typ

Max

Unit

Tx PATH INTERFACE (See Figure 53)

Input Nibble Rate (2Ч Interpolation)

Full

160

MSPS

Input Nibble Rate (4Ч Interpolation)

Full

100

MSPS

Tx Data Setup Time (t

)

Full II

3 ns

Tx Data Hold Time (t

)

Full II

1 ns

Rx PATH INTERFACE (See Figure 54)

Output Nibble Rate

Full

160

MSPS

Rx Data Valid Time (t

)

Full II

3 ns

Rx Data Hold Time (t

)

Full II

0 ns

Explanation of Test Levels

I: 100% production tested.
II: 100% production tested at 25°C and guaranteed by design and characterization at specified temperatures.
III: Sample tested only.
IV: Parameter is guaranteed by design and characterization testing.
V: Parameter is a typical value only.
VI: 100% production tested at 25°C and guaranteed by design and characterization for industrial temperature range.

AD9866

Rev. 0 | Page 9 of 48

ABSOLUTE MAXIMUM RATINGS

Table 8.

Parameter Rating

ELECTRICAL

AVDD, CLKVDD Voltage

3.9 V max

DVDD, DRVDD Voltage

3.9 V max

RX+, RX-, REFT, REFB

-0.3 V to AVDD + 0.3 V

IOUTP+, IOUTP-

-1.5 V to AVDD + 0.3 V

IOUTN+, IOUTN-, IOUTG+,

IOUTG-

-0.3 V to 7 V

OSCIN, XTAL

-0.3 V to CLVDD + 0.3 VS

REFIO, REFADJ

-0.3 V to AVDD + 0.3 V

Digital Input and Output Voltage

-0.3 V to DRVDD + 0.3 V

Digital Output Current

5 mA max

ENVIRONMENTAL

Operating Temperature Range

(Ambient)

-40°C to +85°C

Maximum Junction Temperature

125°C

Lead Temperature (Soldering, 10 s) 150°C
Storage Temperature Range

(Ambient)

-65°C to +150°C

Stresses above those listed under the Absolute Maximum
Ratings may cause permanent damage to the device. This is a
stress rating only; functional operation of the device at these or
any other conditions above those indicated in the operational
section of this specification is not implied. Exposure to absolute
maximum rating conditions for extended periods may affect
device reliability.

THERMAL CHARACTERISTICS

Thermal Resistance: 64-lead LFCSP (4-layer board).

= 24°C/W (paddle soldered to ground plane, 0 LPM air).

= 30.8°C/W (paddle not soldered to ground plane,

0 LPM air).

ESD CAUTION

ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily accumulate on
the human body and test equipment and can discharge without detection. Although this product features
proprietary ESD protection circuitry, permanent damage may occur on devices subjected to high energy
electrostatic discharges. Therefore, proper ESD precautions are recommended to avoid performance
degradation or loss of functionality.

AD9866

Rev. 0 | Page 10 of 48

PIN CONFIGURATION AND FUNCTION DESCRIPTIONS

04560-0-002

19 20

26 27

30 31

62 61

55 54

51 50

ADIO11/Tx[5]

ADIO10/Tx[4]

ADIO9/Tx[3]

ADIO8/Tx[2]

ADIO7/Tx[1]

ADIO6/Tx[0]

ADIO5/Rx[5]

ADIO4/Rx[4]

ADIO3/Rx[3]

ADIO2/Rx[2]

ADIO1/Rx[1]

ADIO0/Rx[0]

RXCLK

TXCLK/TXQUIET

TXEN/TXSYNC

RXEN/RXSYNC

DRV

VSS

CLKOUT1

SEN

GAIN/PGA[

]

PGA

[
4]

PGA

[
3]

PGA

[
2]

PGA

[
1]

VSS

ESET

PGA

[
0]

AVSS

IOUT_N

IOUT_G

AVSS

AVDD

REFIO

REFADJ

AVDD

AVSS

RX+

RX

REFT

AVSS

AVDD

AVSS

DRV

VSS

DWN

CLKOUT2

VSS

CLKV

OSCIN

XTA

CLKV

CONFIG

IOUT_G+

IOUT_N+

IOUT_P

IOUT_P+

AD9866

TOP VIEW

(Not to Scale)

PIN 1
IDENTIFIER

Figure 2. Pin Configuration

Table 9. Pin Function Descriptions

Pin No.

Mnemonic

Mode

Pin Function

ADIO11

MSB of ADIO Buffer

Tx[5]

MSB of Tx Nibble Input

ADIO107

Bits 107 of ADIO Buffer

Tx[41]

Bits 41 of Tx Nibble Input

ADIO6

Bit 6 of ADIO Buffer

Tx[0]

LSB of Tx Nibble Input

ADIO5

Bit 5 of ADIO Buffer

Rx[5]

MSB of Rx Nibble Output

8, 9

ADIO43

Bits 43 of ADIO Buffer

Rx[43]

Bits 43 of Rx Nibble Output

ADIO2

Bit 2 of ADIO Buffer

Rx[2]

Bit 2 of Rx Nibble Output

ADIO1

Bit 1 of ADIO Buffer

Rx[1]

Bit 1 of Rx Nibble Output

ADIO0

LSB of ADIO Buffer

Rx[0]

LSB of Rx Nibble Output

RXEN

ADIO Buffer Control Input

RXSYNC

Rx Data Synchronization Output

TXEN

Tx Path Enable Input

TXSYNC

Tx Data Synchronization Input

TXCLK

ADIO Sample Clock Input

TXQUIET

FD Fast

TxDAC/IAMP

Power-Down

AD9866

Rev. 0 | Page 11 of 48

Pin No.

Mnemonic

Mode

Pin Function

RXCLK

ADIO Request Clock Input

Rx and Tx Clock Output at 2 Ч f

ADC

17, 64

DRVDD

Digital Output Driver Supply Input

18, 63

DRVSS

Digital Output Driver Supply Return

CLKOUT1

DAC

/N Clock Output (L = 1, 2, 4, or 8)

SDIO

Serial Port Data Input/Output

SDO

Serial Port Data Output

SCLK

Serial Port Clock Input

SEN

Serial Port Enable Input

GAIN

Tx Data Port (Tx[5:0]) Mode Select

PGA[5]

HD or FD

MSB of PGA Input Data Port

2529

PGA[40]

HD or FD

Bits 40 of PGA Input Data Port

RESET

Reset Input (Active Low)

31, 34, 36, 39 44, 47, 48

AVSS

Analog Ground

32, 33

REFB, REFT

ADC Reference Decoupling Nodes

35, 40, 43

AVDD

Analog Power Supply Input

37, 38

RX-, RX+

Receive Path - and + Analog Inputs

REFADJ

TxDAC Full-Scale Current Adjust

REFIO

TxDAC Reference Input/Output

IOUT_G-

-Tx Amp Current Output_Sink

IOUT_N-

-Tx Mirror Current Output_Sink

IOUT_G+

+Tx Amp Current Output_Sink

IOUT_N+

+Tx Mirror Current Output_Sink

IOUT_P-

-TxDAC Current Output_Source

IOUT_P+

+TxDAC Current Output_Source

MODE

Digital Interface Mode Select Input
LOW = HD, HIGH = FD

CONFIG

Power-Up SPI Register Default Setting Input

CLKVSS

Clock Osc./Synthesizer Supply Return

XTAL

Crystal Osc. Inverter Output

OSCIN

Crystal Osc. Inverter Input

CLKVDD

Clock Osc./Synthesizer Supply

DVSS

Digital Supply Return

DVDD

Digital Supply Input

CLKOUT2

OSCIN

/L Clock Output, (L = 1, 2, or 4)

PWR_DWN

Power-Down Input

HD = half-duplex mode; FD = full-duplex mode.

AD9866

Rev. 0 | Page 12 of 48

TYPICAL PERFORMANCE CHARACTERISTICS

Rx PATH TYPICAL PERFORMANCE CHARACTERISTICS

AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= f

ADC

= 50 MSPS, low-pass filter's f

-3 dB

= 22 MHz, AIN = -1 dBFS,

RIN = 50 , half- or full-duplex interface, default power bias settings

04560-0-003

FREQUENCY (MHz)

FER

TO IN

T SPEC

(

)

6.25

12.50

18.75

25.00

100

90

80

70

60

50

40

30

20

10

FUND = 1dBFS

SINAD = 61.9dBFS

ENOB = 10BITS

SNR = 64.5dBFS

THD = 65.4dBFS

SFDR = 64.9dBc (THIRD HARMONIC)

RBW = 12.21kHz

Figure 3. Spectral Plot with 4k FFT of Input Sinusoid

with RxPGA = 0 dB and P

= 9 dBm

04493-0-041

FREQUENCY (MHz)

INP

T RE

RRE

D S

CTRUM (dBm)

30

130

120

110

100

90

80

70

60

50

40

RBW = 12.2kHz

Figure 4. Spectral Plot with 4k FFT of 84-Carrier DMT Signal

with PAR = 10.2 dB, P

= -33.7 dBm, and RxPGA = 36 dB

04560-0-005

INPUT AMPLITUDE (dBFS)

(0dBFS = 2V p-p)

I
NAD (dBFS

)

THD (dBFS

)

21

18

15

12

50

92

86

80

74

68

62

56

SINAD @ 3.14V
SINAD @ 3.3V
SINAD @ 3.46V

THD @ 3.14V
THD @ 3.3V
THD @ 3.46V

Figure 5. SINAD and THD vs. Input Amplitude and Supply

= 8 MHz, LPF f

-3 dB

= 26 MHz; Rx PGA = 0 dB)

04560-0-006

RxPGA GAIN (dB)

I
NAD (dBFS

)

ENOB (

10.5

6.5

7.0

7.5

8.0

8.5

9.0

9.5

10.0

1MHz

5MHz

10MHz

15MHz

20MHz

Figure 6. SINAD/ENOB vs. RxPGA Gain and Frequency

04560-0-007

RxPGA GAIN (dB)

THD (dBc

)

55

85

80

75

70

65

60

1MHz

5MHz

10MHz

15MHz

20MHz

Figure 7. THD vs. RxPGA Gain and Frequency

04560-0-008

RxPGA GAIN (dB)

I
NAD (dBFS

)

THD (dBc

)

45

80

75

70

65

60

55

50

SINAD @ +25

SINAD @ +85

SINAD @ 40

THD @ +25

THD @ +85

THD @ 40

Figure 8. SINAD/THD Performance vs. RxPGA Gain

and Temperature ( f

= 5 MHz)

AD9866

Rev. 0 | Page 13 of 48

Rx Path Typical Performance Characteristics:
AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= f

ADC

= 80 MSPS, low-pass filter's f

-3 dB

= 30 MHz, AIN = -1 dBFS,

RIN = 50 , half- or full-duplex interface, default power bias settings

04560-0-009

FREQUENCY (MHz)

FER

TO IN

T SPEC

(

)

100

90

80

70

60

50

40

30

20

10

FUND = 1dBFS

SINAD = 62.4dBFS

ENOB = 10.1BITS

SNR = 63.4dBFS

THD = 69.3dBFS

SFDR = 70.5dBc (THIRD HARMONIC)

RBW = 19.53kHz

Figure 9. Spectral Plot with 4k FFT of Input Sinusoid

with RxPGA = 0 dB and P

= 9 dBm

04560-0-010

FREQUENCY (MHz)

T R

FER

SPEC

(

)

30

130

120

110

100

90

80

70

60

50

40

RBW = 19.53kHz

Figure 10. Spectral Plot with 4K FFT of 111-Carrier DMT Signal

with PAR = 11 dB, P

= -33.7 dBm, LPF's f

-3 dB

= 32 MHz and RxPGA = 36 dB

04560-0-011

INPUT AMPLITUDE (dBFS)

(0dBFS = 2V p-p)

I
NAD (dBFS

)

THD (dBFS

)

21

18

15

12

50

92

86

80

74

68

62

56

SINAD @ 3.14V
SINAD @ 3.3V
SINAD @ 3.46V

THD @ 3.14V
THD @ 3.3V
THD @ 3.46V

Figure 11. SINAD and THD vs. Input Amplitude and Supply

= 8 MHz, LPF f

-3 dB

= 26 MHz; RxPGA = 0 dB)

04560-0-012

RxPGA GAIN (dB)

I
NAD (dBFS

)

ENOB (

10.5

10.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

5MHz

10MHz

15MHz

20MHz

30MHz

Figure 12. SINAD/ENOB vs. RxPGA Gain and Frequency

04560-0-013

RxPGA GAIN (dB)

THD (dBc

)

55

60

65

70

75

80

85

5MHz

10MHz

15MHz

20MHz

30MHz

Figure 13. THD vs. RxPGA Gain and Frequency

04560-0-014

RxPGA GAIN (dB)

I
NAD (dBFS

)

THD (dBc

)

40

80

75

70

65

60

55

50

45

SINAD @ +25

SINAD @ +85

SINAD @ 40

THD @ +25

THD @ +85

THD @ 40

Figure 14. SINAD/THD Performance vs. RxPGA Gain and Temperature

( f

= 10 MHz)

AD9866

Rev. 0 | Page 14 of 48

Rx Path Typical Performance Characteristics:
AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= f

ADC

= 80 MSPS, low-pass filter's f

-3 dB

= 30 MHz, AIN = -1 dBFS,

RIN = 50 , half- or full-duplex interface, default power bias settings

04560-0-015

INPUT FREQUENCY (MHz)

NR (dBFS

)

THD (dBc

)

65.0

60.5

61.0

61.5

62.0

62.5

63.0

63.5

64.0

64.5

60.0

52

54

56

58

60

62

64

66

68

70

72

SNR @ 3.14V
SNR @ 3.3V
SNR @ 3.47V

THD @ 3.14V
THD @ 3.3V
THD @ 3.47V

Figure 15. SNR and THD vs. Input Frequency and Supply

( LPF f

-3 dB

= 26 MHz; RxPGA = 0 dB)

04560-0-016

RxPGA GAIN (dB)

INTE

GRATE

NOIS

(

V rms)

ISE SPEC

ITY (

V/ H

z
)

40

+85

+25

109.4

10.9

21.9

32.8

43.8

54.7

56.6

76.6

87.5

98.5

Figure 16. Input Referred Integrated Noise and Noise Spectral Density

vs. RxPGA Gain (LPF f

-3 dB

= 26 MHz)

04560-0-017

GAIN (dB)

OFFSET (

ll-

scale)

DEVICE 1
DEVICE 2
DEVICE 3
DEVICE 4

Figure 17. Rx DC Offset vs. RxPGA Gain

04560-0-018

INPUT FREQUENCY (MHz)

NR (dBFS

)

THD (dBc

)

20

70

65

60

55

50

45

40

35

30

25

SNR vs. MSPS @ 3.0V

SUP

SNR vs. MSPS @ 346V

SUP

SNR @ 3.13V

THD @ 3.13V
THD @ 3.46V
THD @ 3.3V

Figure 18. SNR and THD vs. Sample Rate and Supply

(LPF Disabled; RxPGA = 0 dB;

= 8 MHz)

04560-0-019

CUTOFF FREQUENCY (MHz)

NR (dB)

Figure 19. SNR vs. Filter Cutoff Frequency

(50 MSPS; f

= 5 MHz; AIN =

-1 dB; RxPGA = 48 dB)

04560-0-020

RxPGA GAIN (dB)

GAIN STEP ERROR (dB)

0.5

0.4

0.3

0.2

0.1

0.2

0.3

0.4

AD9865: GAIN STEP ERROR @ +25

AD9865: GAIN STEP ERROR @ +85

AD9865: GAIN STEP ERROR @ 40

Figure 20. RxPGA Gain Step Error vs. Gain (f

= 10 MHz)

AD9866

Rev. 0 | Page 15 of 48

Rx Path Typical Performance Characteristics:
AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= f

ADC

= 50 MSPS, low-pass filter disabled, RxPGA = 0 dB, AIN = -1 dBFS,

RIN = 50 , half- or full-duplex interface, default power bias settings

04560-0-021

TIME (ns)

CODE

160

240

320

400

480

560

640

720

2048

256

512

768

1024

1280

1536

1792

Figure 21. RxPGA Settling Time -12 dB to +48 dB Transition for DC Input

ADC

= 50 MSPS, LPF Disabled)

04560-0-022

INPUT FREQUENCY (MHz)

PLITU

ESPON

SE (

)

18

15

12

3.3V

3.0V

3.6V

Figure 22. Rx Low-Pass Filter Amplitude Response vs. Supply (f

ADC

= 50 MSPS,

-3 dB

= 33 MHz, RxPGA = 0 dB)

04560-0-023

FREQUENCY (MHz)

ATTEN

@RxPG

(dB)

140

100

110

120

130

TxDAC ISOLATION @ 0dB

IAMP ISOLATION @ 0dB

Figure 23. Rx to Tx Full-Duplex Isolation @ 0 RxPGA Setting
(Note: ATTEN

@ RxPGA = x dB

= ATTEN

@ RxPGA = 0 dB

- RxPGA Gain)

04560-0-024

TIME (ns)

CODE

160

240

320

400

480

560

640

720

1152

1280

1408

256

384

512

640

768

896

1024

Figure 24. RxPGA Settling Time for 0 dB to +5 dB Transition for DC Input

ADC

= 50 MSPS, LPF Disabled)

04560-0-025

INPUT FREQUENCY (MHz)

FUNDAME

TAL (dB)

20

16

18

14

12

10

6dB GAIN
0dB GAIN
+6dB GAIN

+18dB GAIN
+30dB GAIN
+42dB GAIN

Figure 25. Rx Low-Pass Filter Amplitude Response vs. RxPGA Gain

(LPF's f

-3 dB

= 33 MHz)

04493-0-026

FREQUENCY (MHz)

I
S

ANCE

(

)

CAP

ACITANCE

(pF)

105

420

320

330

340

350

360

370

380

390

400

410

Figure 26. Rx Input Impedance vs. Frequency

AD9866

Rev. 0 | Page 16 of 48

TxDAC PATH TYPICAL PERFORMANCE CHARACTERISTICS

AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= 50 MSPS and 80 MSPS, R

SET

= 1.96 k, 2:1 transformer coupled output

(see Figure 63) into 50 load half-or full-duplex interface, default power bias settings

04493-0-072

FREQUENCY (MHz)

dBm

80

70

60

50

40

30

20

10

Figure 27. Dual-Tone Spectral Plot of TxDAC's Output

DATA

= 50 MSPS, 4Ч Interpolation, 10 dBm Peak Power, F1 = 17 MHz,

F2 = 18 MHz)

04560-0-028

2-TONE CENTER FREQUENCY (MHz)

IMD (dBFS

)

(
R

IVE T

PEAK PO

WER)

2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

65

90

85

80

75

70

10dBm

7dBm

4dBm

Figure 28. 2-Tone IMD Frequency Sweep vs. Peak Power

with f

DATA

= 50 MSPS, 4Ч Interpolation

04560-0-029

2-TONE CENTER FREQUENCY (MHz)

SFDR (dBFS)

(
R

IVE T

PEAK PO

WER)

2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

65

90

85

80

75

70

10dBm

7dBm

4dBm

Figure 29. 2-Tone Worst Spur Frequency Sweep vs. Peak Power

with f

DATA

= 50 MSPS, 4Ч Interpolation

04560-0-030

FREQUENCY (MHz)

dBm

80

70

60

50

40

30

20

10

Figure 30. Dual-Tone Spectral Plot of TxDAC's Output

DATA

= 80 MSPS, 2Ч Interpolation, 10 dBm Peak Power, F1 = 27.1 MHz,

F2 = 28.7 MHz)

04560-0-031

2-TONE CENTER FREQUENCY (MHz)

IMD (dBFS)

(
R

IVE T

PEAK PO

WER)

65

90

85

80

75

70

10dBm

7dBm

4dBm

Figure 31. 2-Tone IMD Frequency Sweep vs. Peak Power

with f

DATA

= 80 MSPS, 2Ч Interpolation

04560-0-032

2-TONE CENTER FREQUENCY (MHz)

SFDR (dBFS)

(
R

IVE T

PEAK PO

WER)

65

90

85

80

75

70

10dBm

7dBm

4dBm

Figure 32. 2-Tone Worst Spur Frequency Sweep vs. Peak Power

with f

DATA

= 80 MSPS, 2Ч Interpolation

AD9866

Rev. 0 | Page 17 of 48

TxDAC Path Typical Performance Characteristics:
AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= 50 MSPS and 80 MSPS, R

SET

= 1.96 k, 2:1 transformer coupled output

(see Figure 63) into 50 load, half- or full-duplex interface, default power bias settings

04560-0-033

FREQUENCY (MHz)

dBm

20

30

40

50

60

70

80

90

100

PAR = 11.4

RMS = 1.4

dBm

Figure 33. Spectral Plot of 84-Carrier OFDM Test Vector

DATA

= 50 MSPS, 4Ч Interpolation)

04493-0-079

FREQUENCY (MHz)

dBm

100

125

150

175

200

20

30

40

50

60

70

80

90

100

PAR = 11.4

RMS = 1.4dBm

Figure 34. Wideband Spectral Plot of 88-Subcarrier OFDM Test Vector

DATA

= 50 MSPS, 4Ч Interpolation)

04560-0-035

AOUT (dBFS)

SNR AND 2

-
TONE IMD (dBFS)

(
R

IVE T

PEAK PO

WER)

24

21

18

15

12

105

100

SNR

2-TONE IMD

Figure 35. SNR and SFDR vs. P

OUT

= 12.55 MHz, f

DATA

= 50 MSPS, 4Ч Interpolation)

04493-0-081

FREQUENCY (MHz)

dBm

20

100

90

80

70

60

50

40

30

PAR = 11.4

RMS = 1.4dBm

Figure 36. Spectral Plot of 111-Carrier OFDM Test Vector

DATA

= 80 MSPS, 2Ч Interpolation)

04493-0-082

FREQUENCY (MHz)

dBm

100

120

140

160

20

100

90

80

70

60

50

40

30

PAR = 11.4

RMS = 1.4dBm

Figure 37. Wideband Spectral Plot of 111-Carrier OFDM Test Vector

DATA

= 80 MSPS, 2Ч Interpolation)

04560-0-038

AOUT (dBFS)

SNR AND 2

-
TONE IMD (dBFS)

(
R

IVE T

PEAK PO

WER)

24

21

18

15

12

100

SNR

2-TONE IMD

Figure 38. SNR and SFDR vs. P

OUT

= 20 MHz, f

DATA

= 80 MSPS, 2Ч Interpolation)

AD9866

Rev. 0 | Page 18 of 48

IAMP PATH TYPICAL PERFORMANCE CHARACTERISTICS

AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, f

OSCIN

= 50 MSPS, R

SET

= 1.58 k, 1:1 transformer coupled output (see Figure 64

and Figure 65) into 50 load, half- or full-duplex interface, default power bias settings

04493-0-084

FREQUENCY (MHz)

dBm

60

55

50

45

40

35

30

25

20

15

10

RBW = 2.3kHz

Figure 39. Dual-Tone Spectral Plot of IAMPN Output (IAMP Settings of

I = 12.5 mA, N = 4, G = 0, 2:1 Transformer into 75 Loader, VCM = 4.8 V)

04493-0-085

FREQUENCY (MHz)

dBm

80

70

60

50

40

30

20

10

PAR = 11.4
RMS = 10.3dBm

Figure 40. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMPN in

Current Mode Configuration

(IAMP Settings of I = 10 mA, N = 4, G = 0; VCM = 4.8 V)

04493-0-086

FREQUENCY (MHz)

dBm

80

70

60

50

40

30

20

10

PAR = 11.4
RMS = 10.4dBm

Figure 41. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMP in Voltage

Mode Configuration with AVDD = 5 V

(PBR951 Transistors, IAMP Settings of I = 6 mA, N = 2, G = 6)

04493-0-087

VCM (V)

OIP3

(dBm)

3.0

3.5

4.0

4.5

5.0

5MHz

10MHz

15MHz

20MHz

2.5MHz

Figure 42. IOUTN Third-Order Intercept vs. Common-Mode Voltage

(IAMP Settings of I = 12.5 mA, N = 4, G = 0, 2:1 Transformer into 75 Load)

04493-0-088

VCM (V)

OIP3

(dBm)

3.0

3.5

4.0

4.5

5.0

5MHz

10MHz

15MHz

20MHz

2.5MHz

Figure 43. IOUTG Third-Order Intercept vs. Common-Mode Voltage

(IAMP Settings of I = 4.25 mA, N = 0, G = 6, 2:1 Transformer into 75 Load)

04493-0-089

FREQUENCY (MHz)

dBm

80

70

60

50

40

30

20

10

PAR = 11.4
RMS = 9.8dBm
RBW = 10kHz

Figure 44. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMP in Voltage

Mode Configuration with AVDD = 3.3 V

(PBR951 Transistors, IAMP Settings of I = 6 mA, N = 2, G = 6)

AD9866

Rev. 0 | Page 19 of 48

SERIAL PORT

Table 10. SPI Register Mapping

Power-Up Default Value

MODE = 0 (Half-Duplex)

MODE = 1 (Full-Duplex)

Address
(Hex)

Bit
Break-
down

Description

Width

CONFIG = 0

CONFIG = 1

CONFIG = 0

CONFIG = 1

Comments

SPI PORT CONFIGURATION AND SOFTWARE RESET

0x00

(7)

4-Wire SPI

0 0 0 0

(6)

LSB

First

0 0 0 0

(5)

S/W

Reset

0 0 0 0

Default SPI configuration is
3-wire, MSB first.

POWER CONTROL REGISTERS (via PWR_DWN pin)

0x01

(7)

Clock Syn.

0 0 0 0

(6)

TxDAC/IAMP

0 0 0 0

(5)

Digital

0 0 0 0

(4)

REF

0 0 0 0

(3)

ADC

CML

0 0 0 0

(2)

ADC

0 0 0 0

(1)

PGA

Bias

0 0 0 0

(0)

RxPGA

0 0 0 0

PWR_DWN = 0
Default setting is for all
blocks powered on.

0x02

(7)

CLK Syn.

0 0 0 1*

(6)

TxDAC/IAMP

1 1 1 1

(5)

Digital

1 1 1 1

(4)

REF

1 1 1 1

(3)

ADC

CML

1 1 1 1

(2)

ADC

1 1 1 1

(1)

PGA

Bias

1 1 1 1

(0)

RxPGA

1 1 1 1

PWR_DWN = 1
Default setting* is for all
functional blocks powered
down except PLL.
*MODE = CONFIG = 1
Setting has PLL powered
down with OSCIN input
routed to RXCLK output.

HALF-DUPLEX POWER CONTROL

0x03

(7:3)

Tx OFF Delay

(2)

_TXEN

(1)

PWRDN

(0)

PWRDN

0xFF

N/A N/A

Default setting is for TXEN
input to control power
on/off of Tx/Rx path.
Tx driver delayed by 31
1/f

DATA

clock cycles.

PLL CLOCK MULTIPLIER/SYNTHESIZER CONTROL

0x04

(5)

Duty Cycle Enable

(4) f

ADC

from PLL

1 0

0 0 0

(3:2)

PLL

Divide-N

2 00

00 00 00

(1:0)

PLL

Multiplier-M

2 01

10* 01 01

Default setting is Duty Cycle
Restore disabled, ADC CLK
from OSCIN input, and PLL
multiplier Ч 2 setting. *PLL
multiplier Ч 4 setting.

0x05

(2)

OSCIN to RXCLK

0 0 0 1*

(1)

Invert

RXCLK

0 0 0 0

(0)

Disabled

RXCLK

0 0 0 0

Full-duplex RXCLK normally
at nibble rate.
*Exception on power-up.

0x06

(7:6)

CLKOUT2 Divide

10 10 10 10

(5)

CLKOUT2

Invert

0 0 0 0

(4)

CLKOUT2

Disable

0 0 0 1*

(3:2)

CLKOUT1

Divide

10 10 10 10

(1)

CLKOUT1

Invert

0 0 0 0

(0)

CLKOUT1

Disable

0 0 0 1*

Default setting is CLKOUT2
and CLKOUT1 enabled with
divide-by-2.
*CLKOUT1 and CLKOUT2
disabled.

AD9866

Rev. 0 | Page 20 of 48

Power-Up Default Value

MODE = 0 (Half-Duplex)

MODE = 1 (Full-Duplex)

Address
(Hex)

Bit
Break-
down Description

Width

CONFIG = 0

CONFIG = 1

CONFIG = 0

CONFIG = 1

Comments

Rx PATH CONTROL

0x07

(5)

Initiate Offset Cal.

0 0 0 0

(4)

Rx Low Power

0 1* 0 1*

(0)

Rx Filter ON

1 1 1 1

Default setting has LPF ON
and Rx path at nominal
power bias setting.
*Rx path to low power.

0x08 (7:0)

Rx Filter Tuning
Cut-off Frequency

0x80

0x61*

0x80

Refer to Low-Pass Filter
section.

Tx/Rx PATH GAIN CONTROL

0x09

(6)

Use SPI Rx Gain

(5:0)

Rx Gain Code

0x00

Default setting is for
hardware Rx gain code via
PGA or Tx data port.

0x0A

(6)

Use SPI Tx Gain

(5:0)

Tx Gain Code

0x7F

Default setting is for Tx gain
code via SPI control.

Tx AND Rx PGA CONTROL

0x0B

(6)

PGA Code for Tx

0 0 0 0

(5)

PGA Code for Rx

1 1 1 1

(3)

Force GAIN strobe

0 0 0 0

(2)

Rx Gain on Tx Port

1 0

(1)

3-Bit RxPGA Port

0 1**

0 0

Default setting is RxPGA
control active.
*Tx port with GAIN strobe
(AD9875/AD9876
compatible).
** 3-bit RxPGA gain map
(AD9975 compatible).

Tx DIGITAL FILTER AND INTERFACE

0x0C

(7:6)

Interpolation
Factor

01 00 01 01

(4)

Invert
TXEN/TXSYNC

0 0 0 0

(2)

LS Nibble First*

1 N/A N/A 0

(1)

TXCLK neg. edge

0 0 0 0

(0)

Twos

complement

0 0 1 1

Default setting is 2Ч
interpolation with LPF
response. Data format is
straight binary for half-
duplex and twos
complement for full-duplex
interface.
*Full-duplex only.

Rx INTERFACE AND ANALOG/DIGITAL LOOPBACK

0x0D

(7) Analog

Loopback

0 0 0 0

(6)

Digital

Loopback*

0 0 0 0

(5)

Rx Port 3-State

1 N/A N/A 0

(4)

Invert
RXEN/RXSYNC

0 0 0 0

(2)

LS Nibble First*

1 N/A N/A 0

(1)

RXCLK neg. edge

0 0 0 0

(0)

Twos

complement

0 0 1 1

Data format is straight
binary for half-duplex and
twos complement for full-
duplex interface.
Analog loopback: ADC Rx
data fed back to TxDAC.
Digital loopback: Tx input
data to Rx output port.
*Full-duplex only.

DIGITAL OUTPUT DRIVE STRENGTH, TxDAC OUTPUT, AND REV ID

0x0E

(7)

Low Drive
Strength

0 0 0 0

(0)

TxDAC

Output

0 0 0 0

0x0F

(3:0)

REV ID Number

4 0x00 0x00 0x00 0x00

Default setting is for high
drive strength and IAMP
enabled.

Tx IAMP GAIN AND BIAS CONTROL

0x10

(7)

Select Tx Gain

(6:4)

(2:0)

N 3

0x44

0x44 0x44 0x44

Secondary path G1 = 0, 1, 2,
3, 4.
Primary path N = 0, 1, 2, 3, 4.

0x11

(6:4)

G2 3

(2:0)

G3 3

0x62

0x62 0x62 0x62

Secondary path stages:
G2 = 0 to 1.50 in 0.25 steps
and G3 = 0 to 6.

AD9866

Rev. 0 | Page 21 of 48

Power-Up Default Value

MODE = 0 (Half-Duplex)

MODE = 1 (Full-Duplex)

Address
(Hex)

Bit
Break-
down Description

Width

CONFIG = 0

CONFIG = 1

CONFIG = 0

CONFIG = 1

Comments

0x12

(6:4)

Stand_Secondary

(2:0)

Stand_Primary

0x01

0x01 0x01 0x01

Standing current of primary
and secondary path.

(7:5)

CPGA Bias Adjust

(4:3)

SPGA Bias Adjust

0x13

(2:0)

ADC Bias Adjust

0x00 0x00 0x00 0x00

Current bias setting for Rx
path's functional blocks.
Refer to Page 41.

Bits that are undefined should always be assigned a 0.

REGISTER MAP DESCRIPTION

The AD9866 contains a set of programmable registers described
in Table 10 that can be used to optimize its numerous features,
interface options, and performance parameters from its default
register settings. Registers pertaining to similar functions have
been grouped together and assigned adjacent addresses to
minimize the update time when using the multibyte serial port
interface (SPI) read/write feature. Bits that are undefined within
a register should be assigned a 0 when writing to that register.

The default register settings were intended to allow some
applications to operate without the use of an SPI. The AD9866
can be configured to support a half- or full-duplex digital
interface via the MODE pin with each interface having two
possible default register settings determined by the setting of
the CONFIG pin.

For instance, applications that need to use only the Tx or Rx
path functionality of the AD9866 can configure it for a half-
duplex interface (MODE = 0) and use the TXEN pin to select
between the Tx or Rx signal path with the unused path
remaining in a reduced power state. The CONFIG pin can be
used to select the default interpolation ratio of the Tx path and
RxPGA gain mapping.

SERIAL PORT INTERFACE (SPI)

The serial port of the AD9866 has 3- or 4-wire SPI capability
allowing read/write access to all registers that configure the
device's internal parameters. Registers pertaining to the SPI are
listed in Table 11. The default 3-wire serial communication port
consists of a clock (SCLK), serial port enable (SEN), and a
bidirectional data (SDIO) signal. SEN is an active low control
gating read and write cycles. When SEN is high, SDO and SDIO
are three-stated. The inputs to SCLK, SEN, and SDIO contain a
Schmitt trigger with a nominal hysteresis of 0.4 V centered
about VDDH/2. The SDO pin remains three-stated in a 3-wire
SPI interface.

Table 11. SPI Registers Pertaining to SPI Options

Address (Hex)

Bit

Description

0x00

(7)

Enable 4-wire SPI

(6)

Enable SPI LSB first

A 4-wire SPI can be enabled by setting the 4-wire SPI bit high,
causing the output data to appear on the SDO pin instead of on
the SDIO pin. The SDIO pin serves as an input only, throughout
the read operation. Note that the SDO pin is active only during
the transmission of data and remains three-stated at any other
time.

An 8-bit instruction header must accompany each read and
write operation. The instruction header is shown in Table 12.
The MSB is a R/W indicator bit with logic high indicating a
read operation. The next two bits, N1 and N0, specify the
number of bytes (one to four bytes) to be transferred during the
data transfer cycle. The remaining five bits specify the address
bits to be accessed during the data transfer portion. The data
bits immediately follow the instruction header for both read
and write operations.

Table 12. Instruction Header Information

MSB

LSB

16 15 14 13 12 11 10

R/W

A4 A3 A2 A1 A0

The AD9866 serial port can support both most significant bit
(MSB) first or least significant bit (LSB) first data formats.
Figure 45 illustrates how the serial port words are built for the
MSB first and LSB first modes. The bit order is controlled by the
SPI LSB first bit (Register 0, Bit 6). The default value is 0, MSB
first. Multibyte data transfers in MSB format can be completed
by writing an instruction byte that includes the register address
of the last address to be accessed. The AD9866 automatically
decrements the address for each successive byte required for the
multibyte communication cycle.

AD9866

Rev. 0 | Page 22 of 48

SCLK

SDATA

SCLK

SDATA

R/W N1

D71 D61

D1N D0N

R/W

D01 D11

D7N

D6N

04560-0-045

DATA TRANSFER CYCLE

INSTRUCTION CYCLE

DATA TRANSFER CYCLE

INSTRUCTION CYCLE

SEN

Figure 45. SPI Timing, MSB First (Upper) and LSB First (Lower)

When the SPI LSB first bit is set high, the serial port interprets
both instruction and data bytes LSB first. Multibyte data
transfers in LSB format can be completed by writing an
instruction byte that includes the register address of the first
address to be accessed. The AD9866 automatically increments
the address for each successive byte required for the multibyte
communication cycle.

Figure 46 illustrates the timing requirements for a write
operation to the SPI port. After the serial port enable (SEN)
signal goes low, data (SDIO) pertaining to the instruction
header is read on the rising edges of the clock (SCLK). To
initiate a write operation, the read/not-write bit is set low. After
the instruction header is read, the eight data bits pertaining to
the specified register are shifted into the SDIO pin on the rising
edge of the next eight clock cycles. If a multibyte communica-
tion cycle is specified, the destination address is decremented
(MSB first) and another eight bits of data are shifted in. This
process repeats itself until all the bytes specified in the instruc-
tion header (N1, N0 bits) are shifted in. SEN must remain low
during the data transfer operation, only going high after the last
bit is shifted in.

SEN

SCLK

SDIO

SCLK

LOW

R/W

04560-

046

Figure 46. SPI Write Operation Timing

Figure 47 illustrates the timing for a 3-wire read operation to
the SPI port. After SEN goes low, data (SDIO) pertaining to the
instruction header is read on the rising edges of SCLK. A read
operation occurs if the read/not-write indicator is set high.
After the address bits of the instruction header are read, the
eight data bits pertaining to the specified register are shifted out
of the SDIO pin on the falling edges of the next eight clock
cycles. If a multibyte communication cycle is specified in the
instruction header, a similar process as previously described for
a multibyte SPI write operation applies. The SDO pin remains
three-stated in a 3-wire read operation.

SEN

SCLK

SDIO

SCLK

LOW

R/W

04560-

047

Figure 47. SPI 3-Wire Read Operation Timing

Figure 48 illustrates the timing for a 4-wire read operation to
the SPI port. The timing is similar to the 3-wire read operation
with the exception that data appears at the SDO pin, while the
SDIO pin remains high impedance throughout the operation.
The SDO pin is an active output only during the data transfer
phase and remains three-stated at all other times.

SEN

SCLK

SDIO

SCLK

LOW

R/W

SDO

04560-

048

Figure 48. SPI 4-Wire Read Operation Timing

AD9866

Rev. 0 | Page 23 of 48

DIGITAL INTERFACE

The digital interface port is configurable for half-duplex or full-
duplex operation by pin-strapping the MODE pin low or high,
respectively. In half-duplex mode, the digital interface port
becomes a 10-bit bidirectional bus called the ADIO port. In full-
duplex mode, the digital interface port is divided into two 6-bit
ports called Tx[5:0] and Rx[5:0] for simultaneous Tx and Rx
operations. In this mode, data is transferred between the ASIC
and AD9866 in 6-bit nibbles. The AD9866 also features a
flexible digital interface for updating the RxPGA and TxPGA
gain registers via a 6-bit PGA port or Tx[5:0] port for fast
updates, or via the SPI port for slower updates. See the RXPGA
Control section.

HALF-DUPLEX MODE

The half-duplex mode functions as follows when the MODE
pin is tied low. The bidirectional ADIO port is typically shared
in burst fashion between the transmit path and receive path.
Two control signals, TXEN and RXEN, from a DSP (or digital
ASIC) control the bus direction by enabling the ADIO port's
input latch and output driver, respectively. Two clock signals,
TXCLK and RXCLK, are used to latch the Tx input data and
clock the Rx output data, respectively. The ADIO port can also
be disabled by setting TXEN and RXEN low (default setting),
thus allowing it to be connected to a shared bus.

Internally, the ADIO port consists of an input latch for the Tx
path in parallel with an output latch with three-state outputs for
the Rx path. TXEN is used to enable the input latch; RXEN is
used to three-state the output latch. A five-sample-deep FIFO is
used on the Tx and Rx paths to absorb any phase difference
between the AD9866's internal clocks and the externally
supplied clocks (TXCLK, RXCLK). The ADIO bus accepts input
data-words into the transmit path when the TXEN pin is high,
the RXEN pin is low, and a clock is present on the TXCLK pin,
as shown in Figure 49.

TXCLK

TXEN

ADIO[9:0]

RXEN

TX0

TX2

TX3

TX4

TX1

DIS

04560-0-049

Figure 49. Transmit Data Input Timing Diagram

The Tx interpolation filter(s) following the ADIO port can be
flushed with zeros, if the clock signal into the TXCLK pin is
present for 33 clock cycles after TXEN goes low. Note that the
data on the ADIO bus is irrelevant over this interval.

The output from the receive path is driven onto the ADIO bus
when the RXEN pin is high, and a clock is present on the
RXCLK pin. While the output latch is enabled by RXEN, valid

data appears on the bus after a 6-clock-cycle delay due to the
internal FIFO delay. Note that Rx data is not latched back into
the Tx path, if TXEN is high during this interval with TXCLK
present. The ADIO Bus becomes three-stated once the RXEN
pin returns low. Figure 50 illustrates the receive path output
timing.

PZL

04560-

050

RXEN

ADIO[9:0]

RXCLK

PLZ

RX0

RX1

RX2

RX3

Figure 50. Receive Data Output Timing Diagram

To add flexibility to the digital interface port, several program-
ming options are available in the SPI registers. These options are
listed in Table 13. The default Tx and Rx data input formats are
straight binary, but can be changed to twos complement. The
default TXEN and RXEN settings are active high, but can be set
to opposite polarities, thus allowing them to share the same
control. In this case, the ADIO port can still be placed onto a
shared bus by disabling its input latch via the control signal, and
disabling the output driver via the SPI register. The clock timing
can be independently changed on the transmit and receive
paths by selecting either the rising or falling clock edge as the
validating/sampling edge of the clock. Lastly, the output driver's
strength can be reduced for lower data rate applications.

Table 13. SPI Registers for Half-Duplex Interface

Address (Hex)

Bit Description

0x0C

(4) Invert

TXEN

(1)

TXCLK negative edge

(0)

Twos

complement

0x0D

(5)

Rx port three-state

(4)

Invert

RXEN

(1)

RXCLK negative edge

(0)

Twos

complement

0x0E

(7)

Low digital drive strength

The half-duplex interface can be configured to act like a slave or
a master to the digital ASIC. An example of a slave configura-
tion is shown in Figure 51. In this example, the AD9866 accepts
all the clock and control signals from the digital ASIC. Because
the sampling clocks for the DAC and ADC are derived inter-
nally from the OSCIN signal, it is required that the TXCLK and
RXCLK signals be at exactly the same frequency as the OSCIN
signal. The phase relationships among the TXCLK, RXCLK, and
OSCIN signals can be arbitrary. If the digital ASIC cannot
provide a low jitter clock source to OSCIN, consider using the
AD9866 to generate the clock for its DAC and ADC and pass
the desired clock signal to the digital ASIC via CLKOUT1 or
CLKOUT2.

AD9866

Rev. 0 | Page 24 of 48

TO
Tx DIGITAL
FILTER

ADIO

[11:0]

OSCIN

RXEN

AD9866

FROM
Rx ADC

RXEN

TXEN

TXCLK

RXCLK

DAC_CLK
ADC_CLK

CLKOUT

DIGITAL ASIC

04560-0-051

Tx/Rx

Data[11:0]

Figure 51. Example of a Half -Duplex Digital Interface

with AD9866 Serving as the Slave

Figure 52 shows a half-duplex interface with the AD9866 acting
as the master, generating all the required clocks. CLKOUT1
provides a clock equal to the bus data rate that is fed to the
ASIC as well as back to the TXCLK and RXCLK inputs. This
interface has the advantage of reducing the digital ASIC's pin
count by three. The ASIC needs only to generate a bus control
signal that controls the data flow on the bidirectional bus.

TO
Tx DIGITAL
FILTER

ADIO
[11:0]

Tx/Rx

Data[11:0]

CLKOUT1

AD9866

FROM
Rx ADC

RXEN

TXEN

BUS_CTR

TXCLK

RXCLK

CLKIN

DIGITAL ASIC

04560-0-052

OSCIN

FROM
CRYSTAL
OR MASTER CLK

Figure 52. Example of a Half -Duplex Digital Interface

with AD9866 Serving as the Master

FULL-DUPLEX MODE

The full-duplex mode interface is selected when the MODE pin
is tied high. It can be used for full- or half-duplex applications.
The digital interface port is divided into two 6-bit ports called
Tx[5:0] and Rx[5:0], allowing simultaneous Tx and Rx opera-
tions for full-duplex applications. In half-duplex applications,
the Tx[5:0] port can also be used to provide a fast update of the
RxPGA (AD9876 backward compatible) during an Rx
operation. This feature is enabled by default and can be used to
reduce the required pin count of the ASIC (refer to RxPGA
Control section for more detail).

In either application, Tx and Rx data are transferred between
the ASIC and AD9866 in 6-bit (or 5-bit) nibbles at twice the
internal input/output word rates of the Tx interpolation filter
and ADC. Note that the TxDAC update rate must not be less

than the nibble rate. Therefore, the 2Ч or 4Ч interpolation filter
must be used with a full-duplex interface.

The AD9866 acts as the master, providing RXCLK as an output
clock that is used for the timing of both the Tx[5:0] and Rx[5:0]
ports. RXCLK always runs at the nibble rate and can be inverted
or disabled via an SPI register. Because RXCLK is derived from
the clock synthesizer, it remains active, provided that this
functional block remains powered on. A buffered version of the
signal appearing at OSCIN can also be directed to RXCLK by
setting Bit 2 of Reg. 0x05. This feature allows the AD9866 to be
completely powered down (including the clock synthesizer)
while serving as the master.

The Tx[5:0] port operates in the following manner with the SPI
register default settings. Two consecutive nibbles of the Tx data
are multiplexed together to form a 10-bit data-word in twos
complement format. The clock appearing on the RXCLK pin is
a buffered version of the internal clock used by the Tx[5:0]
port's input latch with a frequency that is always twice the ADC
sample rate (2 Ч f

ADC

). Data from the Tx[5:0] port is read on the

rising edge of this sampling clock, as illustrated in the timing
diagram shown in Figure 53.

Tx 2 LSB

Tx0LSB

RXCLK

TxSYNC

Tx[5:0]

04560-0-053

Tx1MSB

Tx1LSB

Tx2MSB

Tx3LSB

Tx3MSB

Figure 53. Tx[5:0] Port Full-Duplex Timing Diagram

The TXSYNC signal is used to indicate to which word a nibble
belongs. The first nibble of every word is read while TXSYNC is
low as the most significant nibble. The second nibble of that
same word is read on the following TXSYNC high level as the
least significant nibble. If TXSYNC is low for more than one
clock cycle, the last transmit data is read continuously until
TXSYNC is brought high for the second nibble of a new
transmit word. This feature can be used to flush the interpolator
filters with zeros. Note that the GAIN signal must be kept low
during a Tx operation.

The Rx[5:0] port operates in the following manner with the SPI
register default settings. Two consecutive nibbles of the Rx data
are multiplexed together to form a 10-bit data-word in twos
complement format. The Rx data is valid on the rising edge of
RXCLK, as illustrated in the timing diagram shown in
Figure 54. The RXSYNC signal is used to indicate to which
word a nibble belongs. The first nibble of every word is
transmitted while RXSYNC is low as the most significant
nibble. The second nibble of that same word is transmitted on
the following RXSYNC high level as the least significant nibble.

AD9866

Rev. 0 | Page 25 of 48

04560-0-054

RXCLK

RxSYNC

Rx[5:0]

Rx0LSB

Rx1MSB

Rx1LSB

Rx2MSB

Rx3LSB

Rx3MSB

Figure 54. Full-Duplex Rx Port Timing

To add flexibility to the full-duplex digital interface port,
several programming options are available in the SPI registers.
These options are listed in Table 14. The timing for the Tx[5:0]
and/or Rx[5:0] ports can be independently changed by selecting
either the rising or falling clock edge as the sampling/validating
edge of the clock. Inverting RXCLK (via Bit 1 or Reg. 0x0D)
affects both the Rx and Tx interface, because they both use
RXCLK.

Table 14. SPI Registers for Full-Duplex Interface

Address (Hex)

Bit Description

0x05

(2)

OSCIN to RXCLK

(1)

Invert

RXCLK

(0)

Disable

RXCLK

0x0B

(2)

Rx gain on Tx port

0x0C (4)

Invert

TXSYNC

(3)

Tx 5/5 nibble

(2)

LS nibble first

(1)

TXCLK negative edge

(0)

Twos

complement

0x0D

(5)

Rx port three-state

(4)

Invert

RXSYNC

(3)

Rx 5/5 nibble

(2)

LS nibble first

(1)

RXCLK negative edge

(0)

Twos

complement

0x0E

(7)

Low drive strength

The default Tx and Rx data input formats are twos complement,
but can be changed to straight binary. The default TXSYNC and
RXSYNC settings can be changed such that the first nibble of
the word appears while TXSYNC, RXSYNC, or both are high.
Also, the least significant nibble can be selected as the first
nibble of the word (LS nibble first). The output driver strength
can also be reduced for lower data rate applications.

Figure 55 shows a possible digital interface between an ASIC
and the AD9866. The AD9866 serves as the master generating
the required clocks for the ASIC. This interface requires that the
ASIC reserve 16 pins for the interface, assuming a 6-bit nibble
width and the use of the Tx port for RxPGA gain control. Note
that the ASIC pin allocation can be reduced by 3, if a 5-bit
nibble width is used and the gain (or gain strobe) of the RxPGA
is controlled via the SPI port.

04560-0-055

TO
Tx DIGITAL
FILTER

10/12

AD9865/AD9866

FROM
Rx ADC

10/12

RXSYNC

TXSYNC

TX_SYNC

RXCLK

CLKOUT1
CLKOUT2

CLKIN

DIGITAL ASIC

OSCIN

FROM
CRYSTAL
OR MASTER CLK

GAIN

OPTIONAL

Tx Data[5:0]

Rx Data[5:0]

Rx[5:0]

RX_SYNC

MUX

DEMUX

Tx[5:0]

TO
Rx PGA

Figure 55. Example of a Full-Duplex Digital Interface

with Optional RxPGA Gain Control via Tx[5:0]

RxPGA CONTROL

The AD9866 contains a digital PGA in the Rx path that is used
to extend the dynamic range. The RxPGA can be programmed
over a -12 dB to +48 dB with 1 dB resolution using a 6-bit
word, and with a 0 dB setting corresponding to a 2 V p-p input
signal. The 6-bit word is fed into a LUT that is used to distribute
the desired gain over three amplification stages within the Rx
path. Upon power-up, the RxPGA gain register is set to its
minimum gain of -12 dB. The RxPGA gain mapping is shown
in Figure 56. Table 15 lists the SPI registers pertaining to the
RxPGA.

AD9866

Rev. 0 | Page 26 of 48

04560-0-056

6-BIT DIGITAL WORD-DECIMAL EQUIVALENT

GAIN (

12

Figure 56. Digital Gain Mapping of RxPGA

Table 15. SPI Registers RxPGA Control

Address (Hex)

Bit Description

0x09

(6)

Enable RxPGA update via SPI

(5:0)

RxPGA gain code

0x0B

(6)

Select TxPGA via PGA[5:0]

(5)

Select RxPGA via PGA[5:0]

(3)

Enable software GAIN strobe
full-duplex

(2)

Enable RxPGA update via Tx[5:0]
full-duplex

(1)

3-bit RxPGA gain mapping
half-duplex

The RxPGA gain register can be updated via the Tx[5:0] port,
the PGA[5:0] port, or the SPI port. The first two methods allow
fast updates of the RxPGA gain register and should be
considered for digital AGC functions requiring a fast closed-
loop response. The SPI port allows direct update and readback
of the RxPGA gain register via Reg. 0x09 with an update rate
limited to 1.6 MSPS (with SCLK = 32 MHz). Note that Bit 6 of
Reg. 0x09 must be set for a read or write operation.

Updating the RxPGA via the Tx[5:0] port is an option only in
full-duplex mode.

In this case, a high level on the GAIN pin

with Tx SYNC low programs the PGA setting on either the
rising edge or falling edge of RXCLK, as shown in Figure 57.
The GAIN pin must be held high, TxSYNC must be held low,
and GAIN data must be stable for one or more clock cycles to
update the RxPGA gain setting. A low level on the GAIN pin
enables data to be fed to the digital interpolation filter. This
interface should be considered when upgrading existing designs
from the AD9876 MxFE product or half-duplex applications
trying to minimize an ASIC's pin count.

Default setting for full-duplex mode (MODE = 1) .

The GAIN strobe can also be set in software via Reg. 0x0B, Bit 3 for

continuous updating. This eliminates the requirement for external GAIN
signal, reducing the ASIC pin count by 1.

RXCL

SYNC

Tx [5:0

]

GAIN

04560-0-057

Figure 57. Updating RxPGA via Tx[5:0] in Full-Duplex Mode

Updating the RxPGA (or TxPGA) via the PGA[5:0] port is an
option for both the half-duplex

and full-duplex interfaces. The

PGA port consists of an input buffer that passes the 6-bit data
appearing at its input directly to the RxPGA (or TxPGA) gain
register with no gating signal required. Bit 5 or Bit 6 of Reg.
0x0B is used to select whether the data updates the RxPGA or
TxPGA gain register. In applications that switch between
RxPGA and TxPGA gain control via PGA[5:0], be careful that
the RxPGA (or TxPGA) is not inadvertently loaded with the
wrong data during a transition. In the case of an RxPGA to
TxPGA transition, first deselect the RxPGA gain register, update
the PGA[5:0] port with the desired TxPGA gain setting, and
then select the TxPGA gain register.

The RxPGA also offers an alternative 3-bit word gain mapping
option

that provides a -12 dB to +36 dB span in 8 dB incre-

ments as shown in Table 16. The 3-bit word is directed to
PGA[5:3] with PGA[5] being the MSB. This feature is backward
compatible with the AD9975 MxFE and allows direct interfac-
ing to the CX11647 or INT5130 HomePlug 1.0 PHYs.

Table 16. PGA Timing for AD9975 Backward Compatible
Mode

Digital Gain Setting

PGA[5:3]

Decimal

Gain (dB)

000 0

-12

001 1

-12

010

2 -4

011

3 4

100

4 12

101 5

110 6

111 7

Default setting for half-duplex mode (MODE = 0).

Default setting for MODE = 0 and CONFIG =1.

AD9866

Rev. 0 | Page 27 of 48

TxPGA CONTROL

The AD9866 also contains a digital PGA in the Tx path distrib-
uted between the TxDAC and IAMP. The TxPGA is used to
control the peak current from the TxDAC and IAMP over a
7.5 dB and 19.5 dB span, respectively, with 0.5 dB resolution. A
6-bit word is used to set the TxPGA attenuation according to
the mapping shown in Figure 58. The TxDAC gain mapping is
applicable only when Bit 0 of Reg. 0x0E is set, and only the
4 LSBs of the 6-bit gain word are relevant.

04560-0-058

6-BIT DIGITAL CODE (Decimal Equivalent)

ATTE

NUATION (dBFS

)

20

16

18

14

12

10

17

19

15

13

11

TxDACs IOUTP OUTPUT
HAS 7.5dB RANGE

IAMPs IOUTN AND IOUTG
OUTPUTS HAS 19.5dB RANGE

Figure 58. Digital Gain Mapping of TxPGA

The TxPGA register can be updated via the PGA[5:0] port or
SPI port. The first method should be considered for fast updates
of the TxPGA register. Its operation is similar to the description
in the RxPGA Control section. The SPI port allows direct
update and readback of the TxPGA register via Reg. 0x0A with
an update rate limited to 1.6 MSPS (SCLK = 32 MHz). Bit 6 of
Reg 0x0A must be set for a read or write operation. Table 17
lists the SPI registers pertaining to the TxPGA. The TxPGA
control register default setting is for minimum attenuation
(0 dBFS) with the PGA[5:0] port disabled for Tx gain control.

Table 17. SPI Registers TxPGA Control

Address (Hex)

Bit Description

0x0A

(6)

Enable TxPGA update via SPI

(5:0)

TxPGA gain code

0x0B

(6)

Select TxPGA via PGA[5:0]

(5)

Select RxPGA via PGA[5:0]

0x0E

(0)

TxDAC output (IAMP disabled)

AD9866

Rev. 0 | Page 28 of 48

TRANSMIT PATH

The AD9866 (or AD9865) transmit path consists of a selectable
digital 2Ч/4Ч interpolation filter, a 12-bit (or 10-bit) TxDAC,
and a current-output amplifier (IAMP), as shown in Figure 59.
Note that the additional two bits of resolution offered by the
AD9866 (vs. the AD9865) result in a 10 dB to 12 dB reduction
in the pass-band noise floor. The digital interpolation filter
relaxes the Tx analog filtering requirements by simultaneously
reducing the images from the DAC reconstruction process
while increasing the analog filter's transition band. The digital
interpolation filter can also be bypassed, resulting in lower
digital current consumption.

AD9865/AD9866

0 TO 7.5dB

04560-0-059

0 TO 12dB

2-4X

IOUT_G+
IOUT_N+
IOUT_N
IOUT_G

IAMP

IOUT

_P+

IOUT

TXCLK

TXEN/SYNC

ADIO[11:6]/

Tx[5:0]

ADIO[11:6]/

Rx[5:0]

TxDAC

Figure 59. Functional Block Diagram of Tx Path

DIGITAL INTERPOLATION FILTERS

The input data from the Tx port can be fed into a selectable
2Ч/4Ч interpolation filter or directly into the TxDAC (for a half-
duplex only). The interpolation factor for the digital filter is set
via SPI Reg. 0x0C with the settings shown in Table 18. The
maximum input word rate, f

DATA

, into the interpolation filter is

80 MSPS; the maximum DAC update rate is 200 MSPS. There-
fore, applications with input word rates at or below 50 MSPS
can benefit from 4Ч interpolation, while applications with input
word rates between 50 MSPS and 80 MSPS can benefit from
2Ч interpolation.

Table 18. Interpolation Factor Set via SPI Reg. 0x0C

Bits [7:6]

Interpolation Factor

1 (half-duplex only)

Do not use

The interpolation filter consists of two cascaded half-band filter
stages with each stage providing 2Ч interpolation. The first stage
filter consists of 43 taps. The second stage filter, operating at the
higher data rate, consists of 11 taps. The normalized wide band
and pass-band filter responses (relative f

DATA

) for the 2Ч and

4Ч low-pass interpolation filters are shown in Figure 60 and
Figure 61, respectively. Note that these responses also include
the inherent sinc(x) from the TxDAC reconstruction process
and can be used to estimate any post analog filtering
requirements.

The pipeline delays of the 2Ч and 4Ч filter responses are 21.5
and 24 clock cycles, respectively, relative to f

DATA

. The filter delay

is also taken into consideration for applications configured for a
half-duplex interface with the half-duplex power-down mode
enabled. This feature allows the user to set a programmable
delay that powers down the TxDAC and IAMP only after the
last Tx input sample has propagated through the digital filter.
See the Power Control section for more details.

04560-0-060

NORMALIZED FREQUENCY (Relative to

DATA

)

WIDE BAND RESPONSE (dB)

1.25

2.00

90

80

70

60

50

40

30

20

10

1.75

0.75

1.00

1.50

WIDE BAND

0.50

0.25

PASS BAND RESPONSE (dB)

2.5

2.0

1.5

1.0

0.5

1.0

1.5

2.0

PASS BAND

1.0dB @ 0.441

DATA

Figure 60. Frequency Response of 2Ч Interpolation Filter

(Normalized to f

DATA

)

04560-0-061

NORMALIZED FREQUENCY (Relative to

DATA

)

WIDE BAND RESPONSE (dB)

2.5

4.0

90

80

70

60

50

40

30

20

10

3.5

1.5

2.0

3.0

WIDE BAND

1.0

0.5

PASS BAND RESPONSE (dB)

2.5

2.0

1.5

1.0

0.5

1.0

1.5

2.0

PASS BAND

1.0dB @ 0.45

DATA

Figure 61. Frequency Response of 4Ч Interpolation Filter

(Normalized to f

DATA

)

TxDAC AND IAMP ARCHITECTURE

The Tx path contains a TxDAC along with a current amplifier,
IAMP. The TxDAC reconstructs the output of the interpolation
filter and sources a differential current output that can either be
directed to an external load or fed into the IAMP for further
amplification. The TxDAC's and IAMPS's peak current outputs
are digitally programmable over a 0 to -7.5 dB and 0 to -19.5
dB range, respectively, in 0.5 dB increments. Note that this
assumes default register settings for Reg. 0x10 and Reg. 0x11.

AD9866

Rev. 0 | Page 29 of 48

Applications demanding the highest spectral performance
and/or lowest power consumption can use the TxDAC output
directly. The TxDAC is capable of delivering a peak signal
power-up to 10 dBm while maintaining respectable linearity
performance, as shown in Figure 27 through Figure 38. For
power-sensitive applications requiring the highest Tx power
efficiency, the TxDAC's full-scale current output can be reduced
to as low as 2 mA and its load resistors sized to provide a
suitable voltage swing that can be amplified by a low power op-
amp-based driver.

Most applications requiring higher peak signal powers (up to
23 dBm) should consider using the IAMP. The IAMP can be
configured as a current source for loads having a well defined
impedance (50 or 75 systems) or a voltage source (with the
addition of a pair of npn transistors) for poorly defined loads
having varying impedance (such as power lines).

Figure 62 shows the equivalent schematic of the TxDAC and
IAMP. The TxDAC provides a differential current output
appearing at IOUTP+ and IOUTP-. It can be modeled as a
differential current source generating a signal-dependent ac
current, when I

has a peak current of I along with two dc

current sources, sourcing a standing current equal to I. The full-
scale output current, IOUTFS, is equal to the sum of these
standing current sources (IOUTFS = 2*I).

04560-

062

(I+

I
O

UTN

I
O

UTN+

IOU

G+

TxDAC

REFADJ

REFIO

IOUTP+

IOUTP

I +

I 

OFF1

SET

0.1

OFF1

OFF2

IAMP

Figure 62. Equivalent Schematic of TxDAC and IAMP

The value of I is determined by the R

SET

value at the REFADJ

pin along with the Tx path's digital attenuation setting. With
0 dB attenuation, the value of I is

)

(

SET

Equation 1.

For example, an R

SET

value of 1.96 k results in I equal to

10.0 mA with IOUTFS equal to 20.0 mA. Note that the REFIO
pin provides a nominal band gap reference voltage of 1.23 V
and should be decoupled to analog ground via a 0.1 µF
capacitor.

The differential current output of the TxDAC is always con-
nected to the IOUTP pins, but can be directed to the IAMP by
setting Bit 0 of Reg 0x0E. As a result, the IOUTP pins must
remain completely open, if the IAMP is to be used. The IAMP
contains two sets of current mirrors that are used to replicate
the TxDAC's current output with a selectable gain. The first set
of current mirrors is designated as the primary path, providing
a gain factor of N that is programmable from 0 to 4 in steps of 1
via Bits 2:0 of Reg. 0x10 with a default setting of N = 4. Bit 7 of
this register must be set to overwrite the default settings of this
register. This differential path exhibits the best linearity
performance (see Figure 42) and is available at the IOUTN+
and IOUTN- pins. The maximum peak current per output is
100 mA and occurs when the TxDAC's standing current, I, is set
for 12.5 mA (IOUTFS = 25 mA).

The second set of current mirrors is designated as the secon-
dary path providing a gain factor of G that is programmable
from 0 to 36 via Bits 6:4 of Reg. 0x10 and Bits 6:0 of Reg. 0x11
with a default setting of G = 12. This differential path is
intended to be used in the voltage mode configuration to bias
the external npn transistors, because it exhibits degraded
linearity performance (see Figure 43) relative to the primary
path . It is capable of sinking up to 180 mA of peak current into
either its IOUTG+ or IOUTG- pins. The secondary path
actually consists of 3 gain stages (G1, G2, and G3), which are
individually programmable as shown in Table 19. While many
permutations may exist to provide a fixed gain of G, the
linearity performance of a secondary path remains relatively
independent of the various individual gain settings that are
possible to achieve a particular overall gain factor.

Both sets of mirrors sink current, because they originate from
NMOS devices. Therefore, each output pin requires a dc current
path to a positive supply. Although the voltage output of each
output pin can swing between 0.5 and 7 V, optimum ac
performance is typically achieved by limiting the ac voltage
swing with a dc bias voltage set between 4 to 5 V. Lastly, both the
standing current, I, and the ac current, I

, from the TxDAC are

amplified by the gain factor (N and G) with the total standing
current drawn from the positive supply being equal to

)

(

Programmable current sources I

OFF1

and I

OFF2

via Reg. 0x12 can

be used to improve the primary and secondary path mirrors'
linearity performance under certain conditions by increasing
their signal-to-standing current ratio. This feature provides a
marginal improvement in distortion performance under large
signal conditions when the peak ac current of the reconstructed
waveform frequently approaches the dc standing current within
the TxDAC (0 to -1 dBFS sine wave) causing the internal
mirrors to turn off. However, the improvement in distortion
performance diminishes as the crest factor (peak-to-rms ratio)
of the ac signal increases. Most applications can disable these

AD9866

Rev. 0 | Page 30 of 48

current sources (set to 0 mA via Reg. 0x12) to reduce the
IAMP's current consumption.

Table 19. SPI Registers for TxDAC and IAMP

Address (Hex)

Bit

Description

0x0E (0)

TxDAC

output

0x10

(7)

Enable current mirror gain settings

(6:4)

Secondary path first stage gain of 0
to 4 with = 1

(3)

Not used

(2:0)

Primary path NMOS gain of 0 to 4
with = 1

0x11

(7)

Don't care

(6:4)

Secondary path second stage gain of
0 to 1.5 with = 0.25

(3)

Not used

(2:0)

Secondary path third stage gain of 0
to 5 with = 1

0x12

(6:4) IOFF2, secondary path standing

current

(2:0) IOFF1, primary path standing current

Tx PROGRAMMABLE GAIN CONTROL

TxPGA functionality is also available to set the peak output
current from the TxDAC or IAMP. The TxDAC and IAMP are
digitally programmable via the PGA[5:0] port or SPI over a
0 dB to -7.5 dB and 0 dB to -19.5 dB range, respectively, in
0.5 dB increments.

The TxPGA can be considered as two cascaded attenuators with
the TxDAC providing 7.5 dB range in 0.5 dB increments, and
the IAMP providing 12 dB range in 6 dB increments. As a result,
the IAMP's composite 19.5 dB span is valid only if Reg. 0x10
remains at its default setting of 0x44. Modifying this register
setting corrupts the LUT and results in an invalid gain mapping.

TxDAC OUTPUT OPERATION

The differential current output of the TxDAC can be directed to
the IOUTP+ and IOUTP- pins by setting Bit 0 of Reg. 0x0E.
Any load connected to these pins must be ground referenced to
provide a dc path for the current sources. Figure 63 shows the
outputs of the TxDAC driving a doubly terminated 1:1 trans-
former with its center-tap tied to ground. The peak-to-peak
voltage, V p-p,

across R

(and IOUT+ to IOUT-) is equal to

2*I*(R

//R

). With I = 10 mA and R

= R

= 50 , V p-p is equal

to 0.5 V with 1 dBm of peak power being delivered to R

and

1 dBm being dissipated in R

04560-0-063

IOUTN

IOUTN+

IOUTG

IOUTG+

IOU

_
P

IOU

_
P

0 TO 7.5dB

0 TO 12dB

IAMP

FIO

FADJ

SET

0.1

1:1

TxDAC

Figure 63. TxDAC Output Directly via Center-Tap Transformer

The TxDAC is capable of delivering up to 10 dBm peak power
to a load, R

. To increase the peak power for a fixed standing

current, one must increase V p-p across IOUTP+ and
-IOUTP- by increasing one or more of the following parame-
ters: R

, R

(if possible), and/or the turns ratio, N, of transformer.

For example, the removal of R

and the use of a 2:1 impedance

ratio transformer in the previous example results in 10 dBm of
peak power capabilities to the load. Note that increasing the
power output capabilities of the TxDAC reduces the distortion
performance due to the higher voltage swings seen at IOUTP+
and IOUTP-. See Figure 27 through Figure 38 for performance
plots on the TxDAC's ac performance. Optimum distortion
performance can typically be achieved by:

Limiting the peak positive V

IOUTP+

and V

IOUTP

to 0.8 V to

avoid onset of TxDAC's output compression. (TxDAC's
voltage compliance is around 1.2 V.)

Limiting V p-p seen at IOUTP+ and IOUTP- to less than
1.6 V.

Applications demanding higher output voltage swings and
power drive capabilities can benefit from using the IAMP.

IAMP CURRENT MODE OPERATION

The IAMP can be configured for the current mode operation as
shown in Figure 64 for loads remaining relatively constant. In
this mode, the primary path mirrors should be used to deliver
the signal-dependent current to the load via a center-tapped
transformer, because it provides the best linearity performance.
Because the mirrors exhibit a high output impedance, they can
be easily back-terminated (if required).

For peak signal currents (IOUT

up to 50 mA), only the

primary path mirror gain should be used for optimum
distortion performance and power efficiency. The primary
path's gain should be set to 4, with the secondary path's gain
stages set to 0 (Reg. 0x10 = 0x84). The TxDAC's standing
current, I, can be set between 2.5 mA and 12.5 mA with the
IOUTP outputs left open. The IOUTN outputs should be
connected to the transformer, with the IOUTG (and IOUTP)

AD9866

Rev. 0 | Page 31 of 48

outputs left open for optimum linearity performance. The
transformer

should be specified to handle the dc standing

current, I

BIAS

, drawn by the IAMP. Also, because I

BIAS

remains

signal independent, a series resistor (not shown) can be inserted
between AVDD and the transformer's center-tap to reduce the
IAMP's common-mode voltage, V

, and reduce the power

dissipation on the IC. The V

, bias should not exceed 5.0 V and

the power dissipated in the IAMP alone is as follows:

IAMP

)

(

Equation 2.

TxDAC

04560-0-064

IOUTN

IOUTN+

IOUTG

IOUTG+

IOU

_P+

IOU

0 TO 7.5dB

0 TO 12dB

IAMP

REFI

REFADJ

SET

0.1

AVDD

0.1

BIAS

= 2

(N+G)

IOUT

T:1

IOUT

= (N+G)

P_OUT

= (IOUT

)

Figure 64. Current Mode Operation

A step-down transformer

with a turn ratio, T, can be used to

increase the output power, P_OUT, delivered to the load. This
causes the output load, R

, to be reflected back to the IAMP's

differential output by T

resulting in a larger differential voltage

swing seen at the IAMP's output. For example, the IAMP can
deliver 24 dBm of peak power to a 50 load, if a 1.41:1 step-
down transformer is used. This results in 5 V p-p voltage swings
appearing at IOUTN+ and IOUTN- pins. Figure 42 shows how
the third order intercept point, OIP3, of the IAMP varies as a
function of common-mode voltage over a 2.5 MHz to 20.0 MHz
span with a 2-tone signal having a peak power of approximately
24 dBm with IOUT

= 50 mA.

For applications requiring an IOUT

exceeding 50 mA, set the

secondary's path to deliver the additional current to the load.
IOUTG+ and IOUTN+ should be shorted as well as IOUTG-
and IOUTN-. If IOUT

represents the peak current to be

delivered to the load, then the current gain in the secondary
path, G, can be set by the following equation:

IOUT

Equation 3.

The linearity performance becomes limited by the secondary
mirror path's distortion.

The B6080 and BX6090 transformers from Pulse Engineering are worthy of

consideration for current and voltage modes.

IAMP VOLTAGE MODE OPERATION

The voltage mode configuration is shown in Figure 65. This
configuration is suited for applications having a poorly defined
load that can vary over a considerable range. A low impedance
voltage driver can be realized with the addition of two external
RF bipolar npn transistors (Phillips PBR951) and resistors. In
this configuration, the current mirrors in the primary path
(IOUTN outputs) feed into scaling resistors, R, generating a
differential voltage into the bases of the npn transistors. These
transistors are configured as source followers with the secon-
dary path current mirrors appearing at IOUTG+ and IOUTG-
providing a signal-dependent bias current. Note that the IOUTP
outputs must remain open for proper operation.

04560-0-065

IOUTN

IOUTN+

IOUTG

IOUTG+

IOUT_P+

IOUT_P

0 TO 7.5dB

0 TO 12dB

FIO

FADJ

SET

0.1

TO LOAD

AVDD

IOUT

AVDD

0.1

DUAL NPN
PHILLIPS PBR951

IAMP

TxDAC

Figure 65. Voltage Mode Operation

The peak differential voltage signal developed across the npn's
bases is as follows:

)

(

VOUT

Equation 4.

where:

N is the gain setting of the primary mirror.

I is the standing current of the TxDAC defined in Equation 1.

The common-mode bias voltage seen at IOUTN+ and IOUTN-
is approximately AVDD - VOUT

, while the common-mode

voltage seen at IOUTG+ and IOUTG- is approximately the
npn's V

drop below this level (AVDD - VOUT

- 0.65). In

the voltage mode configuration, the total power dissipated
within the IAMP is as follows :

}

)

(

)

{(

VOUT

AVDD

VOUTPK

AVDD

IAMP

Equation 5.

The emitters of the npn transistors are ac-coupled to the
transformer

via a 0.1 µF blocking capacitor and series resistor

of 1 to 2 . Note that protection diodes are not shown for
clarity purposes, but should be considered, if interfacing to a
power or phone line.

AD9866

Rev. 0 | Page 32 of 48

The amount of standing and signal-dependent current used to
bias the npn transistors is dependent on the peak current,
IOUT

, required by the load. If the load is variable, determine

the worst case, IOUT

, and add 3 mA of margin to ensure that

the npn transistors remain in the active region during peak load
currents. The gain of the secondary path, G, and the TxDAC's
standing current, I, can be set using the following equation:

04560-0-066

I (mA)

PPLY

(mA)

100

IAMPN OUTPUT

TxDACs AVDD

IOUT

+ 3

Equation 6.

The voltage output driver exhibits a high output impedance, if
the bias currents for the npn transistors are removed. This
feature is advantageous in half-duplex applications (for example,
power lines) in which the Tx output driver must go into a high
impedance state while in Rx mode. If the AD9866 is configured
for the half-duplex mode (MODE = 0), the IAMP, TxDAC, and
interpolation filter are automatically powered down after a Tx
burst (via TXEN), thus placing the Tx driver into a high
impedance state while reducing its power consumption.

Figure 66. Current Consumption of TxDAC and IAMP in Current Mode

Operation with IOUTN Only (Default IAMP Settings)

04560-0-067

I (mA)

PPLY

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0

100

110

120

130

140

150

IOUTN OUTPUT

IOUTG OUTPUT

TxDAC AVDD

IAMP CURRENT CONSUMPTION CONSIDERATIONS

The Tx path's analog current consumption is an important
consideration when determining its contribution to the overall
on-chip power dissipation. This is especially the case in full-
duplex applications, where the power dissipation can exceed the
maximum limit of 1.66 W, if the IAMP's IOUT

is set to high.

The analog current consumption includes the TxDAC's analog
supply (Pin 43) along with the standing current from the
IAMP's outputs. Equation 2 and Equation 5 can be used to
calculate the power dissipated in the IAMP for the current and
voltage mode configuration. Figure 66 shows the current
consumption for the TxDAC and IAMP as a function of the
TxDAC's standing current, I,

when only the IOUTN outputs are

used. Figure 67 shows the current consumption for the TxDAC
and IAMP as a function of the TxDAC's standing current, I,

when the IOUTN and IOUTG outputs are used. Both figures
are with the default current mirror gain settings of N = 4 and
G = 12.

Figure 67. Current Consumption of TxDAC and IAMP in Current Mode

Operation with IOUTN Only (Default IAMP Settings)

AD9866

Rev. 0 | Page 33 of 48

RECEIVE PATH

The receive path block diagram for the AD9866 (or AD9865) is
shown in Figure 68. The receive signal path consists of a 3-stage
RxPGA, a 3-pole programmable LPF, and a 12-bit (or 10-bit)
ADC. Note that the additional 2 bits of resolution offered by the
AD9866 (vs. the AD9865) result in a 3 dB to 5 dB lower noise
floor depending on the RxPGA gain setting and LPF cutoff
frequency. Also working in conjunction with the receive path is
an offset correction circuit. These blocks are discussed in detail
in the following sections. Note that the power consumption of
the RxPGA can be modified via Reg. 0x13 as discussed in the
Power Control and Dissipation section.

04560-

068

0 TO 6dB

= 1dB

6 TO 18dB

= 6dB

6 TO 24dB

= 6dB

XTAL

RX

10/12

CLK
SYN.

ADC

80MSPS

CLKOUT_1
CLKOUT_2

OSCIN

RX+

CLK

MULTIPLIER

2-POLE

LPF

1-POLE

LPF

SPORT

PGA[5:0]

RXCLK

RXEN/SYNC

ADIO[11:6]/

Tx[5:0]

ADIO[11:6]/

Rx[5:0]

GAIN

MAPPING

LUT

SPGA

AD9865/AD9866

Figure 68. Functional Block Diagram of Rx Path

RX PROGRAMMABLE GAIN AMPLIFIER

The RxPGA has a digitally programmable gain range from
-12 dB to +48 dB with 1 dB resolution via a 6-bit word. Its
purpose is to extend the dynamic range of the Rx path such that
the input of the ADC is presented with a signal that scales
within its fixed 2 V input span. There are multiple ways of
setting the RxPGA's gain as discussed in the RxPGA Control
section, as well as an alternative 3-bit gain mapping having a
range of -12 dB to +36 dB with 8 dB resolution.

The RxPGA is comprised of two sections: a continuous time
PGA (CPGA) for course gain and a switched capacitor PGA
(SPGA) for fine gain resolution. The CPGA consists of two
cascaded gain stages providing a gain range from -12 dB to
+42 dB with 6 dB resolution. The first

stage features a low noise

preamplifier (< 3.0 nV/rtHz), thereby eliminating the need for
an external preamplifier. The SPGA provides a gain range from
0 dB to 6 dB with 1 dB resolution. A look-up table (LUT) is used
to select the appropriate gain setting for each stage.

The nominal differential input impedance of the RxPGA input
appearing at the device RX+ and RX- input pins is 400 //4 pF
(±20%) and remains relatively independent of gain setting. The
PGA input is self-biased at a 1.3 V common-mode level allow-
ing maximum input voltage swings of ±1.5 V at RX+ and RX-.
AC coupling the input signal to this stage via coupling capaci-
tors (0.1 µF) is recommended to ensure that any external dc

offset does not get amplified with high RxPGA gain settings,
potentially exceeding the ADC input range.

To limit the RxPGA's self-induced input offset, an offset
cancellation loop is included. This cancellation loop is
automatically performed upon power-up and can also be
initiated via SPI. During calibration, the RxPGA's first stage is
internally shorted, and each gain stage set to a high gain setting.
A digital servo loop slaves a calibration DAC, which forces the
Rx input offset to be within ±32 LSB for this particular high
gain setting. Although the offset varies for other gain settings,
the offset is typically limited to ±5% of the ADC's 2 V input
span. Note that the offset cancellation circuitry is intended to
reduce the voltage offset attributed to only the RxPGA's input
stage, not any dc offsets attributed to an external source.

The gain of the RxPGA should be set to minimize clipping of
the ADC while utilizing most of its dynamic range. The
maximum peak-to-peak differential voltage that does not result
in clipping of the ADC is shown in Figure 69. While the graph
suggests that maximum input signal for a gain setting of -12 dB
is 8.0 V p-p, the maximum input voltage into the PGA should
be limited to less than 6 V p-p to prevent turning on ESD
protection diodes. For applications having higher maximum
input signals, consider adding an external resistive attenuator
network. While the input sensitivity of the Rx path is degraded
by the amount of attenuation on a dB-to-dB basis, the low noise
characteristics of the RxPGA provide some design margin such
that the external line noise remains the dominant source.

04560-0-069

GAIN (dB)

FULL-S

CALE

AK-TO-P

AK INP

T S

AN (V

)

12

8.0000

4.0000

2.0000

1.0000

0.5000

0.2500

0.1250

0.0625

0.0312

0.0156
0.0100

Figure 69. Maximum Peak-to-Peak Input vs. RxPGA Gain Setting

that Does Not Result in ADC Clipping

AD9866

Rev. 0 | Page 34 of 48

LOW-PASS FILTER

The low-pass filter (LPF) provides a third order response with a
cut-off frequency that is typically programmable over a 15 MHz
to 35 MHz span. Figure 68 shows that the first real pole is
implemented within the first CPGA gain stage, and the complex
pole pair is implemented in the second CPGA gain stage.
Capacitor arrays are used to vary the different R-C time
constants within these two stages in a manner that changes the
cut-off frequency while preserving the normalized frequency
response. Because absolute resistor and capacitor values are
process-dependent, a calibration routine lasting less than 100 µs
automatically occurs each time the target cut-off frequency
register (Reg. 0x08) is updated, ensuring a repeatable cut-off
frequency from device to device.

Although the default setting specifies that the LPF be active, it
can also be bypassed providing a nominal f

-3 dB

of 55 MHz.

Table 20 shows the SPI registers pertaining to the LPF.

Table 20. SPI Registers for Rx Low-Pass Filter

Address (Hex)

Bit

Description

0x07

(0)

Enable Rx LPF

0x08

(7:0)

Target value

The normalized wideband gain response is shown in Figure 70.
The normalized pass-band gain and group delay responses are
shown in Figure 71. The normalized cut-off frequency, f

-3 dB

results in -3 dB attenuation. Also, the actual group delay time
(GDT) response can be calculated given a programmed cut-off
frequency using the following equation:

)

3dB

GDT

Normalized

GDT

Actual

Equation 7.

04560-0-070

FREQUENCY

GAIN (

1.0

3.0

35

30

25

20

15

10

2.5

2.0

1.5

0.5

Figure 70. LPF's Normalized Wideband Gain Response

04560-0-071

NORMALIZED FREQUENCY

GAIN (

0.5

1.0

0.9

0.3

0.4

0.8

0.2

0.1

NORMALIZE

GROUP

LAY

TIM

ESPON

SE (

)

1.30

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1.10

1.15

1.20

1.25

0.25

3.00

2.75

2.50

2.25

2.00

1.75

1.50

1.25

1.00

0.75

0.50

0.25

0.6

0.7

NORMALIZED GROUP DELAY

NORMALIZED GAIN RESPONSE

Figure 71. LPF's Normalized Pass-Band Gain and Group Delay Responses

The -3 dB cut-off frequency, f

-3 dB

, is programmable by writing

an 8-bit word, referred to as the target, to Reg. 0x08. The cut-off
frequency is a function of the ADC sample rate, f

ADC

, and to a

lesser extent RxPGA gain setting (in dB). Figure 72 shows how
the frequency response, f

3 dB

, varies as a function of the RxPGA

gain setting.

04560-0-072

INPUT FREQUENCY (MHz)

FUNDAME

TAL (dB)

18

12

15

6dB GAIN
0dB GAIN
+6dB GAIN
+18dB GAIN
+30dB GAIN
+42dB GAIN

Figure 72. Effects of RxPGA Gain on LPF Frequency Response

( f

-3 dB

= 32 MHz (@ 0 dB and f

ADC

= 80 MSPS)

The following formula

can be used to estimate f

-3 dB

for a

RxPGA gain setting of 0 dB:

)

(

)

(

)

128

(

ADC

target

Equation 8.

Figure 73 compares the measured and calculated f

-3 dB

using this

formula.

Empirically derived for a f

-3 dB

range of 15 MHz to 35 MHz and f

ADC

of 40 MSPS

to 80 MSPS with an RxPGA = 0 dB.

AD9866

Rev. 0 | Page 35 of 48

04560-0-073

TARGET-DECIMAL EQUIVALENT

FRE

NCY

(MHz)

128

224

192

112

176

144 160

208

50 MSPS CALCULATED

80 MSPS CALCULATED

50 MSPS MEASURED

80 MSPS MEASURED

Figure 73. Measured and Calculated f

-3 dB

vs. Target Value

for f

ADC

= 50 MSPS and 80 MSPS

The following scaling factor can be applied to the previous
formula to compensate for the RxPGA gain setting on f

-3 dB

382

)

(

RxPGA

Factor

Scale

Equation 9.

This scaling factor reduces the calculated f

-3 dB

as the RxPGA is

increased. Applications that need to maintain a minimum cut-
off frequency, f

-3 dB_MIN

, for all RxPGA gain settings should first

determine the scaling factor for the highest RxPGA gain setting
to be used. Next, the f

-3 dB_MIN

should be divided by this scale

factor to normalize to the 0 dB RxPGA gain setting (f

-3 dB_0 dB

Equation 8 can then be used to calculate the target value.

The LPF frequency response shows a slight sensitivity to
temperature, as shown in Figure 74. Applications sensitive to
temperature drift can recalibrate the LPF by rewriting the target
value to Reg. 0x08.

04560-0-074

TARGET-DECIMAL EQUIVALENT

FRE

NCY

(MHz)

128

240

192

176

112

144

160

208

OUT

ACTUAL 80MHz AND 40

224

OUT

ACTUAL 80MHz AND +25

OUT

ACTUAL 80MHz AND +85

Figure 74. Temperature Drift of f

-3 dB

for f

ADC

= 80 MSPS and RxPGA = 0 dB

ANALOG TO DIGITAL CONVERTER (ADC)

The AD9866 features a 12-bit analog-to-digital converter
(ADC) capable of up to 80 MSPS. Referring to Figure 68, the
ADC is driven by the SPGA stage, which performs both the
sample-and-hold and the fine gain adjust functions. A buffer
amplifier (not shown) isolates the last CPGA gain stage from
the dynamic load presented by the SPGA stage. The full-scale
input span of the ADC is 2 V p-p, with the full-scale input span
into the SPGA adjustable from 1 V to 2 V in 1 dB increments,
depending on the PGA gain setting.

A pipelined multistage ADC architecture is used to achieve high
sample rates while consuming low power. The ADC distributes
the conversion over several smaller A/D subblocks, refining the
conversion with progressively higher accuracy as it passes the
results from stage to stage on each clock edge. The ADC
typically performs best when driven internally by a 50% duty
cycle clock. This is especially the case when operating the ADC
at high sample rate (55 MSPS to 80 MSPS) and/or lower
internal bias levels, which adversely affect interstage settling
time requirements.

The ADC sampling clock path also includes a duty cycle
restorer circuit, which ensures that the ADC gets a near 50%
duty cycle clock even when presented with a clock source with
poor symmetry (35/65). This circuit should be enabled, if the
ADC sampling clock is a buffered version of the reference
signal appearing at OSCIN (see the Clock Synthesizer section)
and if this reference signal is derived from an oscillator or
crystal whose specified symmetry cannot be guaranteed to be
within 45/55 (or 55/45). This circuit can remain disabled, if the
ADC sampling clock is derived from a divided down version of
the clock synthesizer's VCO, because this clock is near 50%.

The ADC's power consumption can be reduced by 25 mA, with
minimal effect on its performance, by setting Bit 4 of Reg. 0x07.
Alternative power bias settings are also available via Reg. 0x13,
as discussed in the Power Control and Dissipation section.
Lastly, the ADC can be completely powered down for half-
duplex operation, further reducing the AD9866's peak power
consumption.

AD9866

Rev. 0 | Page 36 of 48

04560-0-075

1.0V

ADCs

REFT

REFB

0.1

TOP

VIEW

AGC TIMING CONSIDERATIONS

When implementing a digital AGC timing loop, it is important
to consider the Rx path latency and settling time of the Rx path
in response to a change in gain setting. Figure 21 and Figure 24
show the RxPGA's settling response to a 60 dB and 5 dB change
in gain setting when using the Tx[5:0] or PGA[5:0] port. While
the RxPGA settling time may also show a slight dependency on
the LPF's cutoff frequency, the ADC's pipeline delay along with
the ADIO bus interface presents a more significant delay. The
amount of delay or latency depends on whether a half- or full-
duplex is selected. An impulse response at the RxPGA's input
can be observed after 10.0 ADC clock cycles (1/f

ADC

) in the case

of a half-duplex interface and 10.5 ADC clock cycles in the case
of a full-duplex interface. This latency along with the RxPGA
settling time should be considered to ensure stability of the
AGC loop.

Figure 75. ADC Reference and Decoupling

The ADC has an internal voltage reference and reference ampli-
fier as shown in Figure 75. The internal band gap reference
generates a stable 1 V reference level that is converted to a
differential 1 V reference centered about mid-supply (AVDD/2).
The outputs of the differential reference amplifier are available
at the REFT and REFB pins and must be properly decoupled for
optimum performance. The REFT and REFB pins are conven-
iently situated at the corners of the CSP package such that C1
(0603 type) can be placed directly across its pins. C3 and C4 can
be placed underneath C1, and C2 (10 µF tantalum) can be
placed furthest from the package.

Table 21. SPI Registers for Rx ADC

Address (Hex)

Bit

Description

0x04

(5)

Duty cycle restore circuit

(4)

ADC clock from PLL

0x07

(4)

ADC low power mode

0x13

(2:0)

ADC power bias adjust

AD9866

Rev. 0 | Page 37 of 48

CLOCK SYNTHESIZER

The AD9866 generates all its internal sampling clocks, as well as
two user-programmable clock outputs appearing at CLKOUT1
and CLKOUT2, from a single reference source as shown in
Figure 76. The reference source can be either a fundamental
frequency or an overtone quartz crystal connected between
OSCIN and XTAL with the parallel resonant load components
as specified by the crystal manufacturer. It can also be a TTL-
level clock applied to OSCIN with XTAL left unconnected.

The data rate, f

DATA

, for the Tx and Rx data paths must always be

equal. Therefore, the ADC's sample rate, f

ADC

, is always equal to

DATA

, while the TxDAC update rate is a factor of 1, 2, or 4 of

DATA

, depending on the interpolation factor selected. The data

rate refers to the word rate and should not be confused with the
nibble rate in full-duplex interface.

XTAL

CLK

MULTIPLIER

XTAL

OSCIN

CLKOUT2

CLKOUT1

TO ADC

TO TxDAC

04560-

076

Figure 76. Clock Oscillator and Synthesizer

The 2

CLK multiplier contains a PLL (with integrated loop

filter) and VCO capable of generating an output frequency that
is a multiple of 1, 2, 4, or 8 of its input reference frequency, f

OSCIN

appearing at OSCIN. The input frequency range of f

OSCIN

between 20 MHz and 80 MHz, while the VCO can operate over
a 40 MHz to 200 MHz span. For the best phase noise/jitter
characteristics, it is advisable to operate the VCO with a fre-
quency between 100 MHz and 200 MHz. The VCO output
drives the TxDAC directly such that its update rate, f

DAC

, is

related to f

OSCIN

by the following equation:

OSCIN

DAC

= 2

Equation 10.

where M = 0, 1, 2, or 3.

M is the PLL's multiplication factor set in Reg. 0x04. The value
of M is determined by the Tx path's word rate, f

DATA

, and digital

interpolation factor, F, as shown in the following equation:

)

(

log

OSCIN

DATA

Equation 11.

Note that, if the reference frequency appearing at OSCIN is
chosen to be equal to the AD9866's Tx and Rx path's word rate,
then M is simply equal to log

(F).

The clock source for the ADC can be selected in Reg. 0x04 as a
buffered version of the reference frequency appearing at OSCIN

(default setting) or a divided version of the VCO output (f

DAC

The first option is the default setting and most desirable, if f

OSCIN

is equal to the ADC sample rate, f

ADC

. This option typically

results in the best jitter/phase noise performance for the ADC
sampling clock. The second option is suitable in cases where
f

OSCIN

is a factor of 2 or 4 less than the f

ADC

. In this case, the

divider ratio, N, is chosen such that the divided down VCO
output, f

DAC

, is equal to the ADC sample rate, as shown in the

following equation:

OSCIN

DAC

- )

(

Equation 12.

where N = 0, 1, or 2.

Figure 77 shows the degradation in phase noise performance
imparted onto the ADC's sampling clock for different VCO
output frequencies. In this case, a 25 MHz, 1 V p-p sinewave was
used to drive OSCIN and the PLL's M and N factor were
selected to provide an f

ADC

of 50 MHz for a VCO operating

frequency of 50, 100, and 200 MHz. The RxPGA input was
driven with a near full-scale, 12.5 MHz input signal with a gain
setting of 0 dB. Operating the VCO at the highest possible
frequency results in the best narrow and wideband phase noise
characteristics. For comparison purposes, the clock source for
the ADC was taken directly from OSCIN when driven by a
50 MHz square wave.

04560-0-077

FREQUENCY (MHz)

dBFS

2.5

4.5

6.5

8.5

10.5 12.5 14.5 16.5 18.5 20.5 22.5

110

100

90

80

70

60

50

40

30

20

10

DIRECT
VCO = 50MHz
VCO = 100MHz
VCO = 200MHz

Figure 77. Comparison of Phase Noise Performance when ADC Clock Source

Is Derived from Different VCO Output Frequencies

The CLK synthesizer also has two clock outputs appearing at
CLKOUT1 and CLKOUT2. They are programmable via
Reg. 0x06. Both outputs can be inverted or disabled. The voltage
levels appearing at these outputs are relative to DRVDD and
remain active during a hardware or software reset. Table 22
shows the SPI registers pertaining to the clock synthesizer.

AD9866

Rev. 0 | Page 38 of 48

CLKOUT1 is a divided version of the VCO output and can be
set to be a submultiple integer of f

DAC

, where R = 0, 1, 2,

or 3). Because this clock is actually derived from the same set of
dividers used within the PLL core, it is phase-locked to them
such that its phase relationship relative to the signal appearing
at OSCIN (or RXCLK)can be determined upon power-up. Also,
this clock has near 50% duty cycle, because it is derived from
the VCO. As a result, CLKOUT1 should be selected before
CLKOUT2 as the primary source for system clock distribution.

CLKOUT2 is a divided version of the reference frequency, f

OSCIN

and can be set to be a submultiple integer of f

OSCIN

where L = 0, 1, or 2). With L set to 0, the output of CLKOUT2 is
a delayed version of the signal appearing at OSCIN, exhibiting
the same duty cycle characteristics. With L set to 1 or 2, the
output of CLKOUT2 is a divided version of the OSCIN signal,
exhibiting a near 50% duty cycle, but without having a
deterministic phase relationship relative to CLKOUT1 (or
RXCLK).

Table 22. SPI Registers for CLK Synthesizer

Address (Hex)

Bit

Description

0x04

(4)

ADC CLK from PLL

(3:2)

PLL divide factor ( P)

(1:0)

PLL multiplication factor (M )

0x06

(7:6)

CLKOUT2 divide number

(5)

CLKOUT2 invert

(4)

CLKOUT2 disable

(3:2)

CLKOUT1 divide number

(1)

CLKOUT1 invert

(0)

CLKOUT1 disable

AD9866

Rev. 0 | Page 39 of 48

POWER CONTROL AND DISSIPATION

POWER-DOWN

The AD9866 provides the ability to control the power-on state
of various functional blocks. The state of the PWRDWN pin
along with the contents of Reg. 0x01 and Reg. 0x02 allow two
user-defined power settings that are pin selectable. The default
settings

are such that Reg. 0x01 has all blocks powered on (all

bits 0), while Reg. 0x02 has all blocks powered down excluding
the PLL such that the clock signal remains available at
CLKOUT1 and CLKOUT2. When the PWRDWN pin is low,
the functional blocks corresponding to the bits in Reg. 0x01 are
powered down. When the PWRDWN is high, the functional
blocks corresponding to the bits in Reg. 0x02 are powered
down. PWRDWN immediately affects the designated functional
blocks with minimum digital delay.

Table 23. SPI Registers Associated with Power-Down and
Half-Duplex Power Savings

Address (Hex)

Bit

Description

Comments

0x01

(7) PLL

(6)

TxDAC/IAMP

(5)

Digital

(4)

REF

(3)

ADC

CML

(2)

ADC

(1)

PGA

BIAS

(0)

PGA

PWRDWN = 0

Default setting is all
functional blocks
powered on.

0x02

(7) PLL

(6)

TxDAC/IAMP

(5)

Digital

(4)

REF

(3)

ADC

CML

(2)

ADC

(1)

PGA

BIAS

(0)

PGA

PWRDWN = 1

Default setting is all
functional blocks
powered off
excluding PLL.

0x03

(7:3)

Tx OFF Delay

(2)

Rx PWRDWN
via TXEN

(1)

Enable Tx
PWRDWN

(0)

Enable Rx
PWRDWN

Half-duplex power
savings.

With MODE = 1 and CONFIG = 1, Reg. 0x02 default settings are with all

blocks powered off, with RXCLK providing a buffered version of the signal
appearing at OSCIN. This setting results in the lowest power consumption
upon power-up while still allowing AD9865 to generate the system clock via a
crystal.

HALF-DUPLEX POWER SAVINGS

Significant power savings can be realized in applications having
a half-duplex protocol allowing only the Rx or Tx path to be
operational at any instance. The power-savings method depends
on whether the AD9866 is configured for a full- or half-duplex
interface. Functional blocks having fast power on/off times for
the Tx and Rx path are controlled by the following bits:
TxDAC/IAMP, Tx Digital, ADC, and RxPGA.

In the case of a full-duplex digital interface (MODE = 1), one
can set Reg. 0x01 to 0x60 and Reg. 0x02 to 0x05 (or vice versa)
such that the AD9866's Tx and Rx path are never powered on
simultaneously. The PWRDWN pin can then be used to control
what path is powered on, depending on the burst type. During a
Tx burst, the Rx path's PGA and ADC blocks can typically be
powered down within 100 ns, while the Tx paths DAC, IAMP,
and digital filter blocks are powered up within 0.5 µs. For an Rx
burst, the Tx path's can be powered down within 100 ns, while
the Rx circuitry is powered up within 2 µs.

The TXQUIET pin can also be used with the full-duplex
interface to quickly power down the IAMP and disable the
interpolation filter by setting this pin low. This is meant to
maintain backward compatibility with the AD9875/AD9876
MxFEs with the exception that the TxDAC remains powered if
its IOUTP outputs are used. In most applications, the interpola-
tion filter needs to be flushed with 0s before or after being
powered down. This ensures that, upon power-up, the TxDAC
(and IAMP) have a negligible differential dc offset, thus
preventing spectral splatter due to an impulse transient.

Applications using a half-duplex interface (MODE = 0) can
benefit from an additional power savings feature made available
in Reg. 0x03. This register is effective only for a half-duplex
interface. Besides providing power savings for half-duplex
applications, this feature allows the AD9866 to be used in
applications that need only its Rx (or Tx) path functionality
through pin-strapping, making a serial port interface (SPI)
optional. This feature also allows the PWRDWN pin to retain
its default function as a master power control, as defined in
Table 10.

The default settings for Reg. 0x03 provide fast power control of
the functional blocks in the Tx and Rx signal paths (outlined
above) using the TXEN pin. The TxDAC still remains powered
on in this mode, while the IAMP is powered down. Significant
current savings are typically realized when the IAMP is
powered down.

For a Tx burst, the falling edge of TXEN is used to generate an
internal delayed signal for powering down the Tx circuitry.
Upon receipt of this signal, power-down of the Tx circuitry
occurs within 100 ns. The user-programmable delay for the Tx

AD9866

Rev. 0 | Page 40 of 48

path power-down is meant to match the pipeline delay of the
last Tx burst sample such that power-down of the TxDAC and
IAMP does not impact its transmission. A 5-bit field in Reg.
0x03 sets the delay from 0 to 31 TXCLK clock cycles, with the
default being 31 (0.62 µs with f

TxCLK

= 50 MSPS). The digital

interpolation filter is automatically flushed with midscale
samples prior to power-down, if the clock signal into the
TXCLK pin is present for 33 additional clock cycles after TXEN
returns low. For an Rx burst, the rising edge of TXEN is used to
generate an internal signal (with no delay) that powers up the
Tx circuitry within 0.5 µs.

The Rx path power-on/power-off can be controlled by either
TXEN or RXEN by setting Bit 2 of Reg. 0x03. In the default
setting, the falling edge of TXEN powers up the Rx circuitry
within 2 µs, while the rising edge of TXEN powers down the Rx
circuitry within 0.5 µs. If RXEN is selected as the control signal,
then its rising edge powers up the Rx circuitry and the falling
edge powers it down. It is possible to disable the fast power-
down of the Tx and/or Rx circuitry by setting Bit 1 and/or Bit 0
to 0.

POWER REDUCTION OPTIONS

The power consumption of the AD9866 can be significantly
reduced from its default setting by optimizing the power
consumption versus performance of the various functional
blocks in the Tx and Rx signal path. On the Tx path, minimum
power consumption is realized when the TxDAC output is used
directly and its standing current, I, is reduced to as low as 1 mA.
Although a slight degradation in THD performance results at
reduced standing currents, it often remains adequate for most
applications, because the op amp driver typically limits the
overall linearity performance of the Tx path. The load resistors
used at the TxDAC outputs (IOUTP+ and IOUTP-) can be
increased to generate an adequate differential voltage that can
be further amplified via a power efficient op amp based driver
solution. Figure 78 shows how the supply current for the
TxDAC (Pin 43) is reduced from 55 mA to 14 mA as the
standing current is reduced from 12.5 mA to 1.25 mA. Further
Tx power savings can be achieved by bypassing or reducing the
interpolation factor of the digital filter as shown in Figure 79.

04560-0-078

STANDING

(mA)

IAVDD

TxDAC

(mA)

Figure 78. Reduction in TxDAC's Supply Current vs. Standing Current

04560-0-079

INPUT DATA RATE (MSPS)

DVDD

INTERPOLATION

(HALF-DUPLEX ONLY)

Figure 79. Digital Supply Current Consumption vs. Input Data Rate

(DVDD = DRVDD =3.3 V and f

OUT

= f

DATA

/10)

Power consumption on the Rx path can be achieved by reduc-
ing the bias levels of the various amplifiers contained within the
RxPGA and ADC. As previously noted, the RxPGA consists of
two CPGA amplifiers and one SPGA amplifier. The bias levels
of each of these amplifiers along with the ADC can be con-
trolled via Reg. 0x13 as shown in Table 24. The default setting
for 0x13 is 0x00.

Table 24. SPI Register for RxPGA and ADC Biasing

Address (Hex)

Bit

Description

0x07

(4)

ADC low power

0x13

(7:5)

CPGA bias adjust

(4:3)

SPGA bias adjust

(2:0)

ADC power bias adjust

AD9866

Rev. 0 | Page 41 of 48

Because the CPGA processes signals in the continuous time
domain, its performance versus bias setting remains mostly
independent of the sample rate. Table 25 shows how the typical
current consumption seen at AVDD (Pins 35 and 40) varies as a
function of Bits (7:5), while the remaining bits are maintained at
their default setting of 0. Only four of the possible settings result
in any reduction in current consumption relative to the default
setting. Reducing the bias level typically results in a degradation
in the THD versus frequency performance as shown in
Figure 80 due to a reduction of the amplifier's unity gain
bandwidth, while the SNR performance remains relatively
unaffected.

04560-0-080

CPGA BIAS SETTING-BITS (7:5)

NR (dBFS

)

THD (dBc

)

000

100

010

011

001

65.0

40.0

20

70

65

60

55

50

45

40

35

30

25

42.5

45.0

47.5

50.0

52.5

55.0

57.5

60.0

62.5

SNR_RxPGA = 0dB

SNR_RxPGA = 36dB

THD_RxPGA = 0dB

THD_RxPGA = 36dB

Figure 80. THD vs. f

Performance and RxPGA Bias Settings (000,001,010,100

with RxPGA = 0 and +36 dB and AIN = -1 dBFS, LPF set to 26 MHz

and f

ADC

= 50 MSPS)

Table 25. Analog Supply Current vs. CPGA Bias Settings at
f

ADC

= 65 MSPS

Bit 7

Bit 6

Bit 5

0 0 0 0

0 0 1

-27

0 1 0

-42

0 1 1

-51

1 0 0

-55

1 0 1 27

1 1 0 69

1 1 1 27

The SPGA is implemented as a switched capacitor amplifier.
Therefore, its performance versus bias level is mostly dependent
on the sample rate. Figure 81 shows how the typical current
consumption seen at AVDD (Pins 35 and 40) varies as a
function of bits (4:3) and sample rate, while the remaining bits
are maintained at their default setting of 0. Figure 81 shows how
the SNR and THD performance is affected for a 10 MHz sine
wave input as the ADC sample rate is swept from 20 MHz to 80
MHz.

04560-0-081

ADC SAMPLE RATE (MSPS)

AVDD

(mA)

210

170

175

180

185

190

195

200

205

Figure 81. AVDD Current vs. SPGA Bias Setting and Sample Rate

04560-0-082

SAMPLE RATE (MSPS)

NR (dBc

)

THD (dBc

)

54

74

72

70

68

66

64

62

60

58

56

SNR-00
SNR-01
SNR-10
SNR-11

THD-00

THD-01

THD-10

THD-11

Figure 82. SNR and THD Performance vs. f

ADC

and SPGA Bias Setting with

RxPGA = 0 dB, f

= 10 MHz. AIN =

-1 dBFS

The ADC is based on a pipeline architecture with each stage
consisting of a switched capacitor amplifier. Therefore, its
performance versus bias level is also mostly dependent on the
sample rate. Figure 83 shows how the typical current consump-
tion seen at AVDD (Pins 35 and 40) varies as a function of bits
(2:0) and sample rate, while the remaining bits are maintained
at their default setting of 0. Setting Bit 4 or Reg. 0x07 corre-
sponds to the 011 setting, and the settings of 101 and 111 result
in higher current consumption. Figure 84 shows how the SNR
and THD performance are affected for a 10 MHz sine wave
input for the lower power settings as the ADC sample rate is
swept from 20 MHz to 80 MHz.

AD9866

Rev. 0 | Page 42 of 48

04560-0-083

SAMPLE RATE (MSPS)

IAV

DD (mA)

220

120

130

140

150

160

170

180

190

200

210

000

001

010

011

100

101

101 OR 111

Figure 83. AVDD Current vs. ADC Bias Setting and Sample Rate

04560-0-084

SAMPLE RATE (MSPS)

NR (dBc

)

THD (dBc

)

54

74

72

70

68

66

64

62

60

58

56

THD-000
THD-001
THD-010
THD-011
THD-100
THD-101

SNR-000
SNR-001
SNR-010
SNR-011
SNR-100
SNR-101

Figure 84. SNR and THD Performance vs. f

ADC

and SPGA Bias Setting with

RxPGA = 0 dB, f

= 10 MHz, AIN = -1 dBFS

A sine wave input is a standard and convenient method of
analyzing the performance of a system. However, the amount of
power reduction that is possible is application dependent, based
on the nature of the input waveform (such as frequency content,
peak-to-rms ratio), the minimum ADC sample, and the mini-
mum acceptable level of performance. As a result, it is advisable
that power-sensitive applications optimize the power bias
setting of the Rx path using an input waveform that is repre-
sentative of the application.

POWER DISSIPATION

The power dissipation of the AD9866 can become quite high in
full-duplex applications in which the Tx and Rx paths are
simultaneously operating with nominal power bias settings. In
fact, some applications desiring to use the IAMP may need to
either reduce its peak power capabilities or reduce the power
consumption of the Rx path, so that the device's maximum
allowable power consumption, P

MAX

, is not exceeded.

MAX

is specified at 1.66 W to ensure that the die temperature

does not exceed 125

C at an ambient temperature of 85

C. This

specification is based on the 64-pin LFSCP having a thermal
resistance,

, of 24

C/W with its heat slug soldered. (The

30.8

C/W, if the heat slug remains unsoldered.) If a particular

application's maximum ambient temperature, T

, falls below

C, the maximum allowable power dissipation can be deter-

mined by the following equation:

)

(

MAX

Equation 13.

Assuming that the IAMP's common-mode bias voltage is
operating off the same analog supply as the AD9866, the
following equation can be used to calculate the maximum total
current consumption, I

MAX

, of the IC:

)

(

IAMP

MAX

Equation 14.

With an ambient temperature of up to 85°C, I

MAX

is 478 mA.

If the IAMP is operating off a different supply or in the voltage
mode configuration, first calculate the power dissipated in the
IAMP, P

IAMP

, using Equation 2 or Equation 5, and then recalcu-

late I

MAX

, using the following equation:

)

(

IAMP

MAX

Equation 15.

Figure 78, Figure 79, Figure 81, and Figure 83 can be used to
calculate the current consumption of the Rx and Tx paths for a
given setting.

MODE SELECT UPON POWER-UP AND RESET

The AD9866 power-up state is determined by the logic levels
appearing at the MODE and CONFIG pins. The MODE pin is
used to select a half- or full-duplex interface by pin strapping it
low or high, respectively. The CONFIG pin is used in conjunc-
tion with the MODE pin to determine the default settings for
the SPI registers as outlined in Table 10.

The intent of these particular default settings is to allow some
applications to avoid using the SPI (disabled by pin-strapping
SEN high), thereby reducing the implementation cost. For
example, setting MODE low and CONFIG high configures the
AD9866 to be backward compatible with the AD9975, while
setting MODE high and CONFIG low makes it backward
compatible with the AD9875. Other applications must use the
SPI to configure the device.

A hardware (RESET pin) or software (Bit 5 of Reg. 0x00) reset
can be used to place the AD9866 into a known state of opera-
tion as determined by the state of the MODE and CONFIG
pins. A dc offset calibration and filter tuning routine is also
initiated upon a hardware reset, but not with a software reset.
Neither reset method flushes the digital interpolation filters in
the Tx path. Refer to the Half-Duplex Mode and Full-Duplex
Mode sections for information on flushing the digital filters.

AD9866

Rev. 0 | Page 43 of 48

A hardware reset can be triggered by pulsing the RESET pin low
for a minimum of 50 ns. The SPI registers are instantly reset to
their default settings upon RESET going low, while the dc offset
calibration and filter tuning routine is initiated upon RESET
returning high. To ensure sufficient power-on time of the
various functional blocks, RESET returning high should occur
no less than 10 ms upon power-up. If a digital reset signal from
a microprocessor reset circuit (such as ADM1818) is not
available, a simple R-C network referenced to DVDD can be
used to hold RESET low for approximately 10 ms upon
power-up.

ANALOG AND DIGITAL LOOP-BACK TEST MODES

The AD9866 features analog and digital loop-back capabilities
that can assist in system debug and final test. Analog loop-back
routes the digital output of the ADC back into the Tx data path
prior to the interpolation filters such that the Rx input signal
can be monitored at the output of the TxDAC or IAMP. As a
result, the analog loop-back feature can be used for a half- or
full-duplex interface, to allow testing of the functionality of the
entire IC (excluding the digital data interface).

For example, the user can configure the AD9866 with similar
settings as the target system, inject an input signal (sinusoidal
waveform) into the Rx input, and monitor the quality of the
reconstructed output from the TxDAC or IAMP to ensure a
minimum level of performance. In this test, the user can also
exercise the RxPGA as well as validate the attenuation charac-
teristics of the RxLPF. Note that the RxPGA gain setting should
be selected such that the input does not result in clipping of the
ADC.

Digital loop-back can be used to test the full-duplex digital
interface of the AD9866. In this test, data appearing on the
Tx[5:0] port is routed back to the Rx[5:0] port, thereby
confirming proper bus operation. The Rx port can also be
three-stated for half- and full-duplex interfaces.

Table 26. SPI Registers for Test Modes

Address (Hex)

Bit

Description

0x0D

(7)

Analog loop-back

(6)

Digital loop-back

(5)

Rx port three-state

AD9866

Rev. 0 | Page 44 of 48

PCB DESIGN CONSIDERATIONS

Although the AD9866 is a mixed-signal device, the part should
be treated as an analog component. The on-chip digital cir-
cuitry has been specially designed to minimize the impact of its
digital switching noise on the MxFE's analog performance.

To achieve the best performance, the power, grounding, and
layout recommendations in this section should be followed.
Assembly instructions for the micro-lead frame package can be
found in an application note from Amkor at:
http://www.amkor.com/products/notes_papers/MLF_AppNote
_0902.pdf.

COMPONENT PLACEMENT

If the three following guidelines of component placement are
followed, chances for getting the best performance from the
MxFE are greatly increased. First, manage the path of return
currents flowing in the ground plane so that high frequency
switching currents from the digital circuits do not flow on the
ground plane under the MxFE or analog circuits. Second, keep
noisy digital signal paths and sensitive receive signal paths as
short as possible. Third, keep digital (noise generating) and
analog (noise susceptible) circuits as far away from each other
as possible.

To best manage the return currents, pure digital circuits that
generate high switching currents should be closest to the power
supply entry. This keeps the highest frequency return current
paths short and prevents them from traveling over the sensitive
MxFE and analog portions of the ground plane. Also, these
circuits should be generously bypassed at each device, which
further reduces the high frequency ground currents. The MxFE
should be placed adjacent to the digital circuits, such that the
ground return currents from the digital sections do not flow in
the ground plane under the MxFE.

The AD9866 has several pins that are used to decouple sensitive
internal nodes. These pins are REFIO, REFB, and REFT. The
decoupling capacitors connected to these points should have
low ESR and ESL. These capacitors should be placed as close to
the MxFE as possible (see Figure 75) and be connected directly
to the analog ground plane. The resistor connected to the
REFADJ pin should also be placed close to the device and
connected directly to the analog ground plane.

POWER PLANES AND DECOUPLING

While the AD9866 evaluation board demonstrates a very good
power supply distribution and decoupling strategy, it can be
further simplified for many applications. The board has four
layers: two signal layers, one ground plane, and one power
plane. While the power plane on the evaluation board is split
into multiple analog and digital subsections, a permissible
alternative would be to have AVDD and CLKVDD share the
same analog 3.3 V power plane. A separate analog plane/supply

may be allocated to the IAMP, if its supply voltage differs from
the 3.3 V required by AVDD and CLKVDD. On the digital
side, DVDD and DRVDD can share the same 3.3 V digital
power plane. This digital power plane brings the current used
to power the digital portion of the MxFE and its output drivers.
This digital plane should be kept from going underneath the
analog components.

The analog and digital power planes allocated to the MxFE may
be fed from the same low noise voltage source; however, they
should be decoupled from each other to prevent the noise
generated in the digital portion of the MxFE from corrupting
the AVDD supply. This can be done by using ferrite beads
between the voltage source and the respective analog and
digital power planes with a low ESR, bulk decoupling capacitor
on the MxFE side of the ferrite. Each of the MxFE's supply pins
(AVDD, CLKVDD, DVDD, and DRVDD) should also have a
dedicated low ESR, ESL decoupling capacitors. The decoupling
capacitors should be placed as close to the MxFE supply pins as
possible.

GROUND PLANES

The AD9866 evaluation board uses a single serrated ground
plane to help prevent any high frequency digital ground
currents from coupling over to the analog portion of the
ground plane. The digital currents affiliated with the high speed
data bus interface (Pins 116) have the highest potential of
generating problematic high frequency noise. A ground
serration that contains these currents should reduce the effects
of this potential noise source.

The ground plane directly underneath the MxFE should be
continuous and uniform. The 64-lead LFCSP package is
designed to provide excellent thermal conductivity. This is
partly achieved by incorporating an exposed die paddle on the
bottom surface of the package. However, to take full advantage
of this feature, the PCB must have features to effectively
conduct heat away from the package. This can be achieved by
incorporating thermal pad and thermal vias on the PCB. While
a thermal pad provides a solderable surface on the top surface
of the PCB (to solder the package die paddle on the board),
thermal vias are needed to provide a thermal path to inner
and/or bottom layers of the PCB to remove the heat.

Lastly, all ground connections should be made as short as
possible. This results in the lowest impedance return paths and
the quietest ground connections.

SIGNAL ROUTING

The digital Rx and Tx signal paths should be kept as short as
possible. Also, the impedance of these traces should have a
controlled characteristic impedance of about 50 . This
prevents poor signal integrity and the high currents that can

AD9866

Rev. 0 | Page 46 of 48

EVALUATION BOARD

An evaluation board is available for the AD9865 and AD9866.
The digital interface to the evaluation board can be configured
for a half- or full-duplex interface. Two 40-pin and one 26-pin
male right angle headers (0.100 inches) provide easy interfacing
to test equipment such as digital data capture boards, pattern
generators, or custom digital evaluation boards (FPGA, DSP, or
ASIC). The reference clock source can originate from an
external generator, crystal oscillator, or crystal. Software and an
interface cable are included to allow for programming of the
SPI registers via a PC.

The analog interface on the evaluation board provides a full
analog front-end reference design for power line applications. It
includes a power line socket, line transformer, protection
diodes, and passive filtering components. An auxiliary path
allows independent monitoring of the ac power line. The

evaluation board allows complete optimization of power line
reference designs based around the AD9865 or AD9866.

Alternatively, the evaluation board allows independent
evaluation of the TxDAC, IAMP, and Rx paths via SMA
connectors. The IAMP can be easily configured for a voltage or
current mode interface via jumper settings. The TxDAC's
performance can be evaluated directly or via an optional dual
op amp driver stage. The Rx path includes a transformer and
termination resistor, allowing for a calibrated differential input
signal to be injected into its front end.

More information on the AD9866 evaluation board can be
found at:
http://www.analog.com/Analog_Root/productPage/
productHome/0%2C2121%2CAD9866%2C00.html.

РР»РµРєС‚СЂРѕРЅРЅС‹Р№ РєРѕРјРїРѕРЅРµРЅС‚: AD9866

Document Outline

Р­Р»РµРєС‚СЂРѕРЅРЅС‹Р№ РєРѕРјРїРѕРЅРµРЅС‚: AD9866

Document Outline

РР»РµРєС‚СЂРѕРЅРЅС‹Р№ РєРѕРјРїРѕРЅРµРЅС‚: AD9866